“Seems like the conspiracy theorists are at it again”: conspiracy attributions as a discursive tool in collective (vs. individual) sense-making

Conspiracy theories are often shared, and shaped, through social interaction. Little is known, however, about how these interactions influence and shape people's beliefs. The current research aimed to explore how people reason about the legitimacy of conspiracy theories both privately (individually, by themselves) and publicly (socially, in the presence of others) as a form of individual vs. collective sense-making. In two experiments, participants were presented with information about both the publicly accepted account and a conspiracy account of the origins of COVID-19 (Study 1; N = 244) or vaccinations (Study 2; N = 374). They then deliberated about those accounts either individually by writing a short paragraph reflecting on their thoughts, or publicly by engaging in real-time, unmoderated online group chats with other participants. A Reflexive Thematic Analysis of participant writing (in the form of reflection paragraphs from private deliberations and chat logs from public deliberations) revealed that reasoning broadly centered on the key, structural elements of the theories themselves relating to: plausibility, trust, harm, uncertainty, and autonomy. However, during group deliberation, these themes were employed as discursive tools, influencing the emergent norms for deliberation and approaches to collective sense-making. We find support for the idea that the label “conspiracy theory” is generally negatively perceived and used in everyday social interaction to delegitimize alternative knowledge. We suggest that engaging in open, nuanced dialogues may mitigate backlash effects associated with pejorative use of the “conspiracy” label while empowering people to take an active role in their decision making.

Conspiracy theories are inherently social in their origin and distribution: Individuals believe in conspiracy theories because they have been exposed to them in interpersonal or mass communication. [….] Until recently, the communicative aspects of conspiracy theories have been largely ignored in psychological research. As a result, etiological theories of conspiracy belief may overemphasize individual-level factors over communicative and socio-structural determinants (Douglas and Sutton, 2023, p. 286).

There is a flourishing literature on the psychology of conspiracy theories with now hundreds of studies illuminating the factors that shape endorsement of conspiracy theories (see Biddlestone et al., 2025 and Stasielowicz, 2022 for meta-analyses; Binnendyk and Pennycook, 2022 for a review). Yet, there is growing recognition that conspiracy beliefs, as well as being individually held, are deeply social phenomena (Albarracín, 2022; Douglas and Sutton, 2023). That is, conspiracies are not only beliefs held by people but are transmitted through various forms of communication. Studies have observed that misinformation and conspiracy theories spread rapidly through social media platforms, sometimes at rates exceeding that of scientific or factual content (Cinelli et al., 2022; Vosoughi et al., 2018; Zhang et al., 2021). The social sharing, and shaping of, conspiracy beliefs implicates collective processes whereby people are motivated to share conspiracy theories for impression management (Green et al., 2023), or to build social connections (Ren et al., 2023). We take a different tack to suggest that, in the social sharing of beliefs, people actively rationalize and/or deliberate on the veracity of those claims. That is, they contest the information and make judgments about whether to accept and continue sharing it, or reject and ignore it (Enders et al., 2023b) through a more general process of collective sense-making. These communicative and relational aspects, however, are not yet well understood (Douglas and Sutton, 2023, see above).

We suggest that part of the challenge to addressing the communicative elements lies in the difficulties with defining conspiracy theories themselves. One person's conspiracy is another person's “truth”, and within the literature there is often a lack of clarity in what a conspiracy theory is, and how it may differ from other forms of deceptive content (Mahl et al., 2023). In their recent paper, Douglas and Sutton (2023; see also Nera and Schöpfer, 2023) address this problem by presenting what they term a causal definition of conspiracy theories, focusing on their essential, constitutive properties that conspiracy theories are issues in the public interest but that a) are in opposition to publicly accepted knowledge; b) describe malevolent or forbidden acts; c) ascribe agency to individuals and groups to enact large-scale changes; d) are more prone to falsity than other beliefs (epistemically risky); and e) are social constructs, with the ability not only to describe but also to create social realities (Douglas and Sutton, 2023).

Here, we similarly suggest that the way people rationalize conspiracy theories for (dis)belief is informed by the very properties that define them. That is, we propose that the defining features of conspiracy theories do not just describe them as cognitive structures within people's minds—they actively guide how people reason their way into belief or disbelief (see also Van Den Haak and Grusauskaite, 2025). We therefore ask: how do people subjectively understand and negotiate what is (or is not) a conspiracy theory? In everyday social interactions they emerge from what is already known/accepted, as a counter to these perspectives. When conspiracies are presented against a backdrop of what is already publicly known and accepted, what do people say to justify their position? Drawing on Douglas and Sutton's (2023) framework, we investigate how conspiracy beliefs are constituted through the definitional elements it proposes. Moreover, we extend on this theorizing to address the communicative aspects of that rationalization by considering whether the process differs when deliberating on our beliefs privately (to oneself) vs. publicly (with others). We consider not only how conspiracy beliefs are constituted within the minds of people, but also how an understanding of them evolves within social interactions—processes which have been relatively neglected in the literature to date (Li et al., 2024; see also Van Den Haak and Grusauskaite, 2025).

In two studies, we instructed people to engage with both a conspiratorial and non-conspiratorial account of either the COVID-19 pandemic (Study 1) or vaccination (Study 2) and decide on their own position. Half of the sample deliberated independently, writing a short reflection piece, while the other half did so interactively, in live group chats with 2–3 other participants. We adopt a qualitative approach to explore whether, and how, individual and collective, communicative sense-making relates to the definitional elements outlined by Douglas and Sutton (2023).

Laying theoretical foundations for the study of conspiracy theories

Douglas and Sutton suggest that advancing a causal definition of conspiracy beliefs is critical in allowing the key properties of conspiracy theories to be captured and their “acceptance, sharing, and impacts… inferred” (p. 273). Their analysis identifies key properties that such beliefs are: oppositional, malevolent, agentic, prone to falsity, and socially constructed. These key properties are not static and therefore will not present in the same way and to the same extent in every conspiracy. One implication of such a definition is that if one characteristic is missing, then the belief may not constitute a “conspiracy theory” per se. For example, Harmon-Jones et al. (2024) recent work manipulated the perceived evil of conspirators and observed that high-evil statements were rated as sounding more like a conspiracy by participants than low-evil statements, which moderated the relationship between trait anger and conspiracy beliefs. In the absence of one of the properties (in that case, malevolence), the network of other beliefs may not constitute “a conspiracy theory”.

Importantly, our analysis takes a particularist view (see Dentith, 2016; Hagen et al., 2025) that conspiracies should be evaluated based on their own merit rather than dismissed as a class of flawed beliefs, given that conspiracies do occur (see also Nera and Schöpfer, 2023). This aspect is well captured by the “epistemically risky” characteristic of Douglas and Sutton's (2023) framework, which suggests that, while conspiracies inherently involve secrecy (and, as a result, a lack of an evidentiary trail), powerful people have historically conspired to engage in illegal acts that the public is not privy to (Vermeulen, 2025). Recent scholarship on the language of conspiracies affirms this view, finding that discourse about conspiracy theories (when compared with non-conspiratorial and scientific discourse) is marked by discussions of hierarchies and power asymmetries with a focus on historical abuses of power and cover-ups (Fong et al., 2021; Klein et al., 2019). Here, we explore the subjective experience of encountering and contesting conspiracies, which are presented in the context of (that is, alongside) the publicly accepted versions of events. Given that everyday people make judgements about whether these beliefs are true, valid representations of reality, our focus here is on the content of these judgements. Our first research question therefore asks: how are conspiracies constituted as legitimate (by those who endorse them) or illegitimate (by those who reject them)?

Conspiracy beliefs are individually held and collectively experienced

Conspiracies are specifically those that arise in opposition to publicly accepted versions of events (i.e., they are oppositional; Douglas and Sutton, 2023; Nera and Schöpfer, 2023). It follows that there is a process of comparison and deliberation that occurs when people encounter conspiracy theories which propose alternative explanations for otherwise accepted events. One complexity is that they may do so alone (individually, through private reflection) but also interactively (socially, in discussions that they have with others). Conspiracies can make people feel uncertain and, when uncertain, people are often motivated to discuss their beliefs with others (e.g., Hogg, 2021). Research indeed finds that people discuss, share (Cinelli et al., 2022) and sometimes act on their conspiracy beliefs to achieve social and political ends (Hornsey and Pearson, 2022; Thomas et al., 2024). At the outbreak of the COVID-19 pandemic, for instance, online interactions surged as people shared information about the outbreak, its spread and prevention strategies (e.g., Park et al., 2021). Thus, alongside private internal deliberation about these events, there was also an active process of collective information sharing and sense-making.

Notably, both internal reasoning and “public” discussions are likely to be shaped by pre-existing identities and ideologies. People internalize their relatedness with others through group memberships and, as such, even private reflection is not disconnected from one's social identities or realities (Turner et al., 1987). The context may differ (i.e., deliberation can occur independently or in groups), but their individual approaches to problem-solving as well as their social representations can be present in both. A priori ideologies are also likely to be a key resource that would be drawn up on as people reason about conspiracies, publicly or privately (see Jost and Amodio, 2012).

Ideology, then, represents the lens through which we structure and interpret the world around us (Forberg, 2023). As such, people may be particularly drawn to discuss conspiracies when they resonate with their existing political, social, or cultural values. For example, political partisans may endorse conspiracies which depict their own party as a moral, “valued ingroup” (van Prooijen and Song, 2021, p. 456) targeted by a vilified representation of their political opposition (see also Douglas et al., 2019; Sternisko et al., 2020). An explicit focus on the communicative aspects of conspiracy belief formation invites consideration of the ways in which such beliefs are contested (e.g., Nera et al., 2020) and socially influenced.

In the context of collective sense-making, though, judgments about conspiracies cannot simply be reduced to the sum of individual beliefs. Rather, emergent group processes are likely to shape the approaches to and outcomes of such collective meaning-making. Groups are vehicles for social influence whereby members are invited to reflect on both the unique (unshared) information they hold and the information that has been shared amongst the group (Stasser and Davis, 1981). How members respond to this information, what they choose to prioritize and the associated norms that are developed (i.e., norms for critical thinking or consensus; Postmes et al., 2001) therefore determine the outcomes of group interaction, with consequences for beliefs and behavior (Borinca et al., 2024). Cognition can therefore become socially shared, characterized as a “collaborative process that occurs between individuals” (Levine, 2018, p. 52). As such, it is necessary to attend both to the intrapersonal reasoning and communicative aspects of belief formation. Given that conspiracy theories hold relevance to people as individuals and as group members, our next research question asks: how do people make sense of conspiracy theories in private vs. explicitly social contexts? We do not assume a priori that the distinction between these two modes of deliberation is contained specifically to one factor (i.e., individual reasoning, norms, etc.), but rather that it is potentially attributable to all of these factors.

The current research

The current research explores how conspiracy theories are rationalized—or rejected—when people reflect on them individually (privately; through an internal process of reasoning and deliberation) vs. collectively (publicly; through an interactive process of collective reasoning and deliberation). Van Den Haak and Grusauskaite (2025, p. 2) recently advocated for a stronger focus on how people “situate, narrate and justify” changes in their beliefs. Psychological investigations of conspiracy beliefs are overwhelmingly cross-sectional and/or laboratory-based, conducted using quantitative methods in which a standard approach is to present participants with a series of statements and then average across them as a measure of conspiracy belief (Nera et al., 2020). Therefore, the central theoretical contribution of the current paper is in the exploration of how structural features of conspiracy theories are mobilized as discursive tools in social interaction, and how group norms shape collective sense-making around contested claims. Our methods allow us to draw comparisons between individual vs. collective sense-making, inviting a more qualitative, inductive approach (which is relatively under-utilized in this literature) alongside the more explicit framework provided by Douglas and Sutton (2023). There are, of course, rich sources of observational data available through social media (Cinelli et al., 2022; Zhang et al., 2021; see also Zollo et al., 2015). However, people who are drawn to interact online about their beliefs are likely to be those who are already more certain (Enders et al., 2023b) than those who are exposed and need to decide what to make of them. Accordingly, we overcome these dual challenges by introducing a more general sample to conspiratorial explanations for events against the backdrop of the publicly accepted versions and ask them to deliberate and reach a decision about their dis/belief either by themselves or in a real time online group chat with other participants. The content of the reflections/chat logs is therefore our primary corpus.

We conducted two studies where we introduced participants to two accounts of issues that were in the public interest: a COVID-19 conspiracy (Study 1) and a vaccination conspiracy (Study 2) via written excerpts. Study 1 was conducted in mid-2021 during the roll out of the first COVID-19 vaccines in the United States while new variants caused infection cases to continue rising; Study 2 was conducted in early 2022 during the introduction of booster vaccinations and growing population immunity in the US. There was still rapid viral spread, high rates of hospitalization and extensive societal discussion and debate about the origins of the disease (Study 1) and the role of vaccination (Study 2) at the time these data were collected. While the pandemic is a unique recent incident, it is also the case that conspiracy theories flourish during societal crises (Uscinski and Parent, 2014; van Prooijen and Douglas, 2017; Fong et al., 2021). The present study adopts conspiracy beliefs about the origins of COVID-19 and vaccination as a test case in which to explore more general themes of individual vs. collective sense-making. Importantly, the stimuli were designed to introduce both the conspiratorial explanation and publicly accepted explanation of events. We then instructed participants to deliberate on the stimuli to reach a decision about whether they agreed or disagreed with the conspiratorial interpretation (or were unable to reach a decision). Critically, those in the group deliberation condition were instructed to reach a decision with others, via a 10-min live text-based chat with 2–3 other participants, while those in the individual deliberation condition were instructed to reach a decision independently by writing a short reflection.

As we have discussed above, deliberating independently does not imply that participants are necessarily disconnected from their social identities and realities. However, it is important that these modes of deliberation still be treated as conceptually distinct. Where individual deliberation involves drawing on one's own pre-existing notions, deliberating interactively involves the company of others who may need to be convinced of a given viewpoint to reach a unanimous decision. These are fundamentally different experiences and yet are composite of how people arrive at and consolidate their beliefs. The data therefore form a rich and ecologically valid account of judgments about conspiracy theories, which are well placed to inform our aims of inferring the key properties of conspiracy beliefs according to those who actually interact with them. As a result, we may also identify novel patterns and relationships that could further inform theoretical development.

We adopt the approach of Reflexive Thematic Analysis (Braun et al., 2023), which acknowledges the active role the researcher plays in the production of knowledge. Braun and Clarke (2019) note that this approach is intended to offer researchers flexibility regarding how theory informs its implementation, and caution against binaries (i.e., deductive or inductive). As such, we apply a more deductive theoretical framework to the data to generate codes relating to each of the five definitional characteristics of conspiracy theories (see Douglas and Sutton, 2023). Supplementary Table 1 - adapted from Kibiswa's (2019) coding frame diagram – in the Supplementary materials contains a summary of each of the components along with operational definitions adapted from Douglas and Sutton (2023) to guide the coding process. The list of deductive codes generated from the theoretical framework are used to structure and categorize emergent patterns and themes from the data. These codes are then used inductively, enabling us to consider the meaning participants communicated through their responses (Braun et al., 2023) and allowing for further codes to be created. Responses could be coded against more than one category if necessary. Codes were quantified (as count data) as an analytic aid for qualitative interpretation and understanding. We do not assume that codes represent objective truth per se but rather include frequency counts to elucidate patterns and to further inform our interpretive analysis, especially in relation to the comparative aspects of our research questions.

We adopted a critical realist epistemology, which suggests that reality exists at three levels: empirical (what we can experience or observe), actual (events that occur, whether we observe them or not), and real (the underlying mechanisms that cause events to occur) (Fletcher, 2017). That is, while the real social world exists outside of our knowledge, we can attempt to understand it by using theory to identify mechanisms and make inferences about their effects on our empirical observations.

The data were collected as part of a larger project exploring the interaction between person factors and social influence processes on evaluations of conspiracy theories (Study 1 https://osf.io/tm4an?mode=&revisionId=&view_only=75a96203575f475cb8718a8e9496d9f and Study 2 https://osf.io/q25sz?view_only=008ea1df95934d1e958aba1dbe00472b registrations are accessible on OSF). Two experimental studies were conducted, using a paradigm that tested the endorsement of specific conspiracy beliefs when the theories were deliberated on within groups relative to individuals. Data was collected via Amazon Mechanical Turk, which was logistically necessary to enable the recruitment of large samples of participants to populate multiple, simultaneous chat rooms. The design resulted in quantitative survey data (which are not examined here), as well as qualitative data, based on the written deliberations and group chat logs. To analyze this data, we implemented the six-phase approach to Reflexive Thematic Analysis developed by Braun et al. (2023). As such, we (1) familiarized ourselves with the data, which involved an initial read of subsets of the textual data; (2) coded the data to capture features relevant to the research questions; (3) generated initial themes by identifying patterns between codes as they related to a core concept; (4) reviewed and developed themes to ensure they adequately represented the data; (5) refined, defined, and named the themes through discussion between the first and second authors; and (6) began producing the analysis.

Participants

Pandemic conspiracy

Participants for Study 1 (N = 244) were residents of the United States recruited primarily through Amazon Mechanical Turk in August 2021. Half of the participants were male (52.0%, 46.7% female, 0.8% non-binary, 0.4% did not specify), aged 18 or older (M = 36.7, SD = 11.70), and the majority were moderately liberal (31.6%, compared to 23.4% very liberal, 16.0% neither liberal nor conservative, 11.1% moderately conservative, 7.0% center liberal, 7.4% center conservative, 3.7% very conservative).

Vaccination conspiracy

Study 2 used the same paradigm and design as Study 1 and was collected in February 2022. Participants (N = 374) were residents of the United States recruited through Amazon Mechanical Turk. Additionally, half were female (51.1%, 47.1% male, 1.6% non-binary, 0.3% did not specify) and moderately liberal (22.7%, compared to 20.1% very liberal, 15.8% neither, 15.5% center liberal, 10.2% center conservative, 10.2% moderate conservative, 5.6% very conservative).

Procedure

The materials and approach were identical for the two samples, with items being worded to refer to the relevant issue (a pandemic conspiracy in Study 1, and a vaccination conspiracy in Study 2). Participants completed pre-measures and were instructed to read a short excerpt about the conspiracy that “powerful elites are behind the COVID-19 Pandemic” (Study 1) or that “pharmaceutical companies are faking data on vaccine efficacy and safety” (Study 2). Critically, the stimuli contained balanced information, first presenting the conspiratorial account which was outside/opposed to public knowledge, and then presenting a counter argument consisting of publicly available/accepted knowledge. Participants were then randomly allocated to either the individual (Study 1, n = 114; Study 2, n = 175) or group (Study 1, n = 130; Study 2, n = 199) deliberation condition. To ensure that randomization between the two conditions was successful, we conducted independent samples t -tests on age, gender, and political orientation. There were no significant differences for Study 1 nor Study 2; Supplementary Tables 2, 3 displays the results.

Participants in the individual condition responded to a prompt that read:

“Please reflect on the ideas presented to you a moment ago. That is, that wealthy elites (Dr. Fauci, Bill Gates) are using their power and the virus to profit from vaccines [that vaccinations are unsafe and have unreported adverse effects, and that they are encouraged to help make pharmaceutical companies wealthy]. We are particularly interested in whether you agree or disagree with this assertion and ask that you reach a decision about whether you oppose or support this belief and why you have adopted this position.”

Participants responded to this open-ended question with a time limit of 7 min and no word limit. Responses to this question form the basis of our analysis at the individual level. Participants wrote M = 126.61, SD = 62.60 words for Study 1 and M = 128.93, SD = 58.66 words for Study 2.

Participants in the interaction condition read the following prompt:

“As a group, please reflect on the ideas presented to you a moment ago. That is, that wealthy elites (Dr. Fauci, Bill Gates) are using their power and the virus to profit from vaccines [that vaccinations are unsafe and have unreported adverse effects, and that they are encouraged to help make pharmaceutical companies wealthy]. We are particularly interested in whether you agree or disagree with this assertion. We ask that you reach a group decision about whether you oppose or support this belief and why you have adopted this position.”

Participants then engaged in live, unmoderated group chats with 2–3 other participants with a time limit of 10 min. and no limit on the number of messages or words within the messages. We judged that a chat duration of less than 10 min would not provide sufficient time for a unanimous decision to be reached. Given that group deliberation also required participants to first coordinate on the task and share their opinion (tasks not necessary in the individual condition), we allowed more time in the group (vs. individual) deliberation condition. We also judged that it was important to avoid introducing the confound of boredom for people in the individual condition by requiring them to spend 10 min doing a task which could be completed in 7 min. The average number of words within the chats (cumulatively, accounting for the writing of all participants within a group) was M = 391.85, SD = 116.30 for Study 1 and M = 463.00, SD = 142.72 for Study 2, from M = 37.06, SD = 8.90 messages per group sent in Study 1 and M = 39.16, SD = 15.91 in Study 2. Following the manipulation, all participants were asked to report whether they agreed or disagreed with the conspiracy, or whether they were unable to reach a decision. Additionally, participants were instructed to rate their familiarity with each conspiracy, from (1) not at all familiar to (6) extremely familiar (Study 1, M = 3.30, SD = 1.45; Study 2, M = 3.91, SD = 1.39). Finally, all participants were debriefed and were referred to fact checking articles regarding the origins and spread of COVID-19 at the completion of Study 1 and the World Health Organization's vaccine information series at the completion of Study 2.

By way of our positionality, we note that we approached the coding as primarily quantitative researchers, and our analytic decisions and structuring of results are strongly influenced by our training in quantitative psychology with its accompanying positivist epistemology. However, both researchers are trained in the meta theoretical perspective of the social identity approach which recognizes the ways in which social categories are deployed to make meaning of social reality—a key meta theoretical underpinning of this analysis. Moreover, we recognize our position as researchers employed in a higher education institution (a university) and we paid reflexive attention to the inherent power relations as academic researchers who study people who may have less institutional power, and our role as arbiters in deciding what is “credible” vs. what is “conspiracy”. These positions made us attentive to how aspects of our identities, as well as those of participants, inform knowledge production.

The initial coding was conducted using NVivo (Lumivero, 2023) by the first author. Individual responses for Study 1 were read first, followed by individual responses for Study 2. Group chat transcripts for Study 1 were then read, followed by transcripts for Study 2. Once coding was completed, the first and second authors discussed and considered how these categories related to each other to extract higher-order themes. We then reviewed each theme to ensure that they accurately represented the data. In keeping with outlines for using thematic analysis specified by Braun and Clarke (2006), the significance of any given theme is not necessarily determined by how prevalent it is in the data, but by how relevant it is to the research questions. We compared categories and themes across the two studies (pandemic conspiracy vs. vaccination conspiracy) as well as across deliberative conditions (individual vs. group deliberation) to examine similarities and nuances. Table 1 contains a map of the themes and sub-themes (both those derived from the framework and those which emerged from the data), how often they occurred, and illustrative examples.

Participant quotes are identified with S1 or S2 to denote study number, followed by a participant number for those in the individual deliberation condition. Those in the group deliberation condition are identified by their group number (G), as well as their User number. Overall, five central themes emerged that reflect how people deliberate about the “truth” (or not) of conspiracy theories. Table 1 shows that three of these were consistently present across both Studies 1 and 2, and two emerged only within the context of the vaccination conspiracy (Study 2). Chi-square tests were conducted for each sub-theme to explore patterns of recurrence. We do not treat frequency as a marker of importance, however, but rather to aid in our construction of themes and interpretation of how they map onto deliberation conditions. We focus here on our qualitative interpretations - the results of this more “deductive” approach are displayed in Supplementary Table 4. Figures 1, 2 display a map of the themes and sub-themes, across the individual/group deliberation conditions and across opinions on the conspiracy (agree/disagree, based on cross-tabulation comparisons).

The results revealed that conspiracy theories are interpreted (in the literature, as well as in the “wild”) through consistent key features (see Table 1). Three higher-order themes that represented these features were prevalent across both studies (plausibility, trust, and harm). Plausibility reflects participants' efforts to evaluate the evidence for and against the conspiracy in the context of public knowledge. That is, a series of sub-themes emerged relating to the idea that conspiracies lack an evidentiary trace and are constructed in opposition of what is commonly known or accepted (per the “epistemically risky” and “oppositional” and “socially constructed” factors; Douglas and Sutton, 2023). Trust refers mainly to participants' perceptions of the alleged conspirators – whether they are regarded positively or negatively, the degree to which they hold power (economic, social etc.) and their capability, willingness, or intention to engage in the conspiracy (per the “agency” factor). Harm relates to the outcomes of the conspiracy, the associated actions of the alleged conspirators and whether their motivations are “malicious” (S1-G31, User 3) or “altruistic” (S2-G44, User 3) (per the “describes malevolent or forbidden acts” factor). The two other higher-order themes occurred only in the context of Study 2 (vaccination): these were uncertainty and autonomy. The uncertainty theme suggested that, while participants may not have had grievances with vaccination prior to the pandemic, they felt that information about COVID-19 vaccinations was insufficient. As a result, some participants believed that these vaccinations were potentially unsafe. Finally, the autonomy theme suggested that participants felt their sense of control/self-determination was violated or challenged regarding COVID-19 vaccination mandates, which sometimes led to vaccine hesitancy. We explore how these themes play out between studies and across modes of deliberation in the analysis that follows.

What is (or, is not) a conspiracy? conspiracy beliefs as a discursive tool

Figures 1, 2 show that many sub-themes were represented in participant deliberations in both contexts (COVID-19 origins and vaccination), across deliberation conditions and position on conspiracy. The figures also highlight, however, that sub-themes are skewed toward “disagree”, a result of the fact that most participants rejected both the COVID-19 and vaccination conspiracies (n disagree = 419, n agree = 84, n unsure = 115). These results suggest that, in general, people recognize and reason about key, structural features of conspiracy theories that align with Douglas and Sutton's (2023) conceptualization; agreement or disagreement, then, depends on how these elements are construed.

For example, participants consistently identified the agency and power that was afforded to the alleged conspirators by their wealth. This power invited suspicion, and participants noted the potential for corruption or disregard for the wellbeing of others in the pursuit of maintaining that power. Crucially, however, participants who rejected the theories explicitly negated these elements:

User 2: The government needs to regulate vaccines to make sure they are relatively safe and effective.

User 3: The vaccine is safe and the government is just funding research and development so that it is safe

User 4: That is what I believe as well, as the vaccinations have shown to reduce the possibility and dangers of the viruses.

User 1: I feel like a vaccine is the best tool we have to combat illnesses. Of course there are going to be side effects and it will effect each person differently. I think the government does what it can to investigate those side effects and will approve vaccines if in 99.9% of cases, the side effects are relatively mild.

User 4: Do you guys think that pharmaceutical industries are getting a profit from these vaccination pushes?

User 2: Of course they do. That doesn't mean there's some nefarious purpose behind them. Profit motive creates competition and leads to creativity.

User 1: Pharmaceutical industries are making a profit, but someone has to make a profit otherwise there would be no incentive to come up with new medicines/vaccines

...

User 4: That is true, but there are also cases [where] the companies make a profit to the point in which they monopolize how the medicine and treatments are distributed. This, in turn, affects the greater good, and is seen with the battles against STDs remedies.

...

User 2: That's why we need to both invest in oversight agencies and create transparency around the government to lobbying cycle.

(S2–G5)

Here, participants initiate the exchange by interactively reinforcing the idea that vaccine safety is regularly monitored, implicitly establishing a norm for trust in authorities (e.g., science and government/pharmaceutical institutions). This norm sets the tone for the deliberation, wherein users draw on their pre-existing notions about wealth and power, acknowledging the potential for structural harm caused by perverse incentives (“pharmaceutical companies are making a profit”, “of course they do”, “there are cases [where] the companies make a profit to the point in which they monopolize how the medicine and treatments are distributed”). However, users ultimately revert to their normative expectation that pharmaceutical industries can be trusted, explicitly negating these harms by discussing the role of regulation and transparency in mitigating these risks, countering the tendency to overestimate the capabilities of agents implicated in conspiracies (Douglas and Sutton, 2023). Discounting this structural feature of a conspiracy by establishing a norm for trust early in the discussion appeared to curb conspiracism, with G5 ultimately rejecting the conspiratorial account, despite their apparent skepticism. Similarly, other participants who rejected the conspiracies implied that, while some authorities might have intentions to mislead or harm, the conditions required for the plots to be successful were extreme enough to undermine their plausibility (Grimes, 2016). For example, one participant commented:

It's not even that it's such an outlandish idea that rich and powerful people would do nefarious things to benefit themselves, it's just that these particular theories are usually so completely illogical… For these theories to be true it would take a global effort of scientists and governments of every major country working in unison, otherwise everything would be readily debunked. Not only does it seem kind of insane to assume the entirety of the worlds powers are in on this conspiracy, but for that number of people to also keep quiet about it for this long would just be impossible.

(S1–P61)

Interestingly, P61 also explicitly labels the alternative account as a conspiracy theory. Table 1 highlights that in Study 1, people interpreted the alternative account of the origins and spread of COVID-19 as a “clear and obvious conspiracy” (S1-P16), whereas in Study 2, participants did so less than half of the time. The above excerpt illustrates that when participants perceived the claims as being implausible or “illogical” (i.e., epistemically risky), as more malevolent or “nefarious”, and as exaggerating the capabilities of the conspirators (“to also keep quiet about it for this long would just be impossible”), the claims are interpreted as “conspiracy-like” (Harmon-Jones et al., 2024; Raab et al., 2013) and, often, rejected (per Figures 1, 2).

In contrast, participants who accepted the conspiratorial accounts did not think it impossible or unlikely that such a scale of “evil” could exist. Similarly to the above, these participants enlist their existing notions about wealth and power as a rhetorical strategy to inform their current beliefs. These participants primarily perceived that the information received from experts about the pandemic and vaccines was falsified as the result of an intentional, coordinated attempt to mislead (as is displayed in Figures 1, 2). Indeed, people who hold conspiracy beliefs tend to view the social world as non-random and a “bad place that conspires against them”, where individual agents lack a fixed set of moral standards (Moulding et al., 2016, p. 353), reflecting dangerous worldviews (see Biddlestone et al., 2025). For example:

I agree that it's plausible, and I'd argue likely but such sinister intentions are much easier dismissed by people who wish to think the world doesn't have such evil in it.. it's easier to label such people making suggestions as this, as “conspiracy theorists” but I say “why not?” At least some of whom mentioned here are clearly motivated by wealth, therefore it would be greatly to their financial benefit, at break-neck speeds, to push and exaggerate an opportunity like a global pandemic...Clearly anyone who challenges or cites studies or science to contradict what is force-fed to the public is quickly shutdown and censored

(S1-P43)

Here, P43 suggests that there is a malevolent, harmful agenda that is orchestrated by a wealthy group. Rather than drawing on positive experiences in which perverse incentives are challenged (per the previous exchange between users in G5), P43 draws on negative experiences (for example, of censorship) as a rhetorical strategy, commenting that they felt they were not “allowed” to have a different opinion. This excerpt also highlights explicit awareness that acceptance of the belief is contrary to the publicly accepted version of events. Their deliberation therefore reflects on the challenges of raising alternatives that “contradict what is force-fed to the public”, indicating that conspiracy beliefs can also be a form of reactance against the implied censorship of adhering to the accepted versions of events (Biddlestone et al., 2021). This participant's view that bad actors are already conspiring is reinforced by their perception that alternative viewpoints are suppressed; such a threat to their freedom of expression then seems to result in the defensive rejection of mainstream scientific communication. Indeed, recent theorizing suggests that censorship as an approach to managing mis/disinformation could backfire, interpreted as being carried out by the conspirators to keep the plot hidden from the public (Albarracín, 2022).

As well as citing malevolent plots to misrepresent or mislead the public, some participants cited inconsistencies in scientific communication as justification for their endorsement of the conspiracy belief (see also Jennings et al., 2021):

User 2: That... makes him [Dr Anthony Facui] seem a bit more nefarious if he's getting money from books on this.

User 4: He said we shouldn't wear masks, then he said we should, then he said double masking, then back to single. He was never very consistent.

User 3: yeah there seem to be many reasons why it is hard to judge his true motivations

(S1-G7)

Here, participants frame scientific uncertainty and evolving recommendations as evidence of dishonesty, which acts to reinforce the alleged malevolent agenda in the context of the origins of COVID-19. Interpreting inconsistency as evidence of corruption or coordinated acts of malice rather than as a feature of scientific progress aligns with research suggesting that a lack of scientific reasoning skills hinders people's ability to navigate complex information (Georgiou et al., 2021, 2023).

The alleged conspirator also emerged as a factor of each theory that influenced how participants evaluated the conspiratorial accounts. In Study 1, specific individuals (Bill Gates and Dr. Anthony Fauci) were implicated, whereas in Study 2, the conspirator was an institution (Pharmaceutical companies). Several participants (see Table 1) noted that they maintained a personal dislike for the people implicated in the COVID-19 conspiracy account, critiquing their personal lives and character. For some, these critiques served to reduce trust in the “elites” in question. For example, S1-G76, User 2 commented that Gates is “ Just a greedy person helping out his rich friends.” The conspirator therefore plays an important role in people's evaluations of conspiracy theories: moral blame for undesired outcomes can be placed on those specific individuals who are implicated, while the systemic causes are more easily ignored (Jolley et al., 2018). As we have also illustrated above with the exchange between participants in Group 5, when the conspiracy implicated an institution, they tended to explain the structural reasons behind the alleged outcomes and propose structural solutions, rather than focusing on individual blame. For example:

“(I'm sure there IS a ton of profit in these vaccines, though. But that's capitalism. Ideally, we make doing the right thing profitable for the companies ” (S2-G17, User 3).

Individual vs. collective sense-making

A secondary part of our analysis was to examine how the process of deliberation and rationalization differed when people did it alone, vs. publicly (through interactions with others). Figures 1, 2 imply that the mode of deliberation interacted with the Study/topic (COVID-19, vaccination) to shape the outcomes. As we have discussed above, the label “conspiracy” was applied much more frequently in Study 1 compared to Study 2. Intriguingly, however the use of the term “conspiracy theory” in the context of the group deliberation had downstream consequences for the tone and content of subsequent discussions. Participants across both deliberation conditions made humorous, disparaging remarks about each of the conspiracies and those who endorse them (e.g., S1–P89: “ I understand that people can be dumb as hell, but it's insane to me that people actually believe this ”; S1–G28, User 3: “ there are variants of stupidity, just like covid”). However, Figure 1 suggests that in Study 1, participants almost exclusively engaged in this behavior in the group deliberation condition. Figure 2, on the other hand, reveals that participants in Study 2 made similar humorous, disparaging comments predominantly in the individual deliberation condition.

The following exchange highlights the tendency in Study 1 for participants to rapidly reach a consensus that claims about Bill Gates and Dr. Fauci's perverse incentives in creating and spreading the coronavirus are explicitly conspiratorial (and, by extension, false):

User 1: so i have no reason to think there's a giant conspiracy lol. no evidence

User 3: I personally don't believe that there is a conspiracy.

User 4: I agree with both of you. No conspiracy.

User 2: Me neither.

User 1: i can invent any conspiracy i want to

User 1: donald trump is responsible

User 1: obama is responsible

User 2: As long as you put it online! LOL.

User 1: either there's a giant conspiracy orchestrated across the world's leaders and scientists... or it's just another virus lol

(S1-G9)

In this exchange, User 1 initiates the discussion by immediately labeling the alternative account as a conspiracy and sharing that they do not believe any evidence for it exists. This notion is sequentially repeated by the remaining group members, reinforcing its legitimacy and, ultimately, making the statement more influential (Stasser and Davis, 1981). Group members rapidly converged around their shared opinion, and this consensus appears to result in a shared “anti-conspiracy” identity (Borinca et al., 2024). Users created a norm for ridicule in this context, employing satire to highlight the absurdity of the conspiracy. In this instance, participants verbal behavior serves not only to further delegitimize the claim but to protect their ingroup image by strengthening the divide between “us” and “them” (“them” being highly irrational conspiracy theorists; Orosz et al., 2016). Within Study 1, participants in group chats with similar trajectories were often unwilling to be publicly perceived as supporting claims that were (even tangentially) related to the conspiratorial account, with any form of dissent being swiftly suppressed:

User 2: seems like the conspiracy theorist are at it again

...

User 1: dr faci isn't even super rich to be getting money from it

User 2: True.. But being the devils advocate I did watch some documentary where he had some linkage to vaccines

User 2: but can't really recollect the exact connection

User 4: It is just hype from conspiracy theorists. I don't see a real link between any of it.

(S1-G6)

Again, “conspiracy theorists” are represented as “others” outside the rational boundaries of this (small) deliberating group. Such a construction from Users 2 and 4 would effectively quell contributions from others that might have suspected aspects of the claims—or even their own “devils advocacy”. While there is relatively little research exploring how the label “conspiracy theory” is generally perceived (Nera et al., 2020), there is some evidence to suggest it has a negative connotation and may be used to stigmatize those who hold them (Lantian et al., 2018). Indeed, use of the label can even become weaponised, employed to negate the legitimacy of the claim (Nera et al., 2020). Given that the conspiratorial account was explicitly called out as such and often ridiculed in Study 1 group discussions, participant writing reflected these negative perceptions. Rather than critically assessing the validity of the claims, participants dismissed them because they appeared “conspiratorial” or because they originated from individuals labeled as “conspiracy theorists”, potentially reflecting a Type 2 error (Vermeulen, 2025).

By contrast, the group deliberation had rather different content in Study 2 and displayed more openness to diverse perspectives, potentially because participants were far less likely to stigmatize alternative claims by labeling them as an outright conspiracy (see Table 1). Rather, some participants noted that elements of the conspiratorial account might even be true:

User 4: I can definitely understand the perspective that vaccines are being pushed for profit. I'm not so certain about the claim that they are exactly “unsafe” though.

User 3: I agree with that. I think there needs to be more of a separation between the science and scientists who developed these vaccines and the companies who end up selling them.

User 1: I think they are probably generally safe, but some will and do have bad reactions. So, that's why it should be left to each person to decide based on their likelihood to have a reaction (previous reactions) and their likelihood of getting very sick from Covid.

User 1: I think most are honestly more concerned about future side effects, especially with the mrna vaccines.

User 1: Most immediate side effects are pretty short lived and minor.

User 2: Right, but we have to weigh that against possible future side effects of covid, which is also an unknown.

User 1: Very true. Which is why I think it should be up to individuals to decide.

(S2-G8)

Although claims that vaccines are promoted for profit despite adverse effects are linked to the alternative, conspiratorial account (rather than the publicly accepted account), participants in this group did not explicitly frame them as a conspiracy. Unlike the exchange between S1–G6 where “conspiracy-like” information was suppressed, here participants treat each piece of information as equally valid by affirming one another's perspectives (“I agree with that”; “Right”; “Very true”). Rather than publicly distancing themselves from this belief, these participants adopted a more nuanced stance, accepting only some aspects of the claim (“I can definitely understand the perspective that vaccines are being pushed for profit”) while rejecting others (“I'm not so certain about the claim that they are exactly “unsafe” though”).

Consequently, Figures 1, 2 show that themes unique to Study 2 (uncertainty and autonomy) emerged, and almost exclusively in the group deliberation condition. These discussions were centered around the concerns that people had about a vaccine that (at the time) was entirely new to them:

User 3: I usually think they [vaccines] are good. Sometimes I get paranoid.

User 4: So I am fully vaccinated and boosted. I do not believe that there are hidden side effects. There are some adverse effects in an extremely small portion of people who are vaccinated.

User 2: I was paranoid at first with the covid vaccine too. There is nothing wrong with not being first in line lol

User 4: That's understandable considering how fast it was developed

(S2-G42)

Participants in another chat commented:

User 1: I think most are honestly more concerned about future side effects, especially with the mrna vaccines.

User 1: Most immediate side effects are pretty short lived and minor.

User 2: Right, but we have to weigh that against possible future side effects of covid, which is also an unknown.

User 1: Very true. Which is why I think it should be up to individuals to decide.

User 4: There is a complete lack of long term data on mrna vaccines, that is something I totally agree with and it makes me worried

(S2-G8)

Figure 2 shows that these themes were highly common among participants whose position was “unsure”. These groups evaluated the legitimacy of both the publicly accepted and the conspiratorial accounts of vaccine safety and efficacy and were often unable to unanimously reach a decision to accept or reject the conspiracy. Groups that could not unanimously reach a decision were characterized by norms that permitted expressions of uncertainty, allowing concerns, dissenting opinions and skepticism to be voiced without being treated as an explicit endorsement of the conspiratorial explanation (“I was paranoid…too. There is nothing wrong with not being first in line”). The norms for negotiation within these groups allowed for a more nuanced exploration of the broader discussion around vaccine safety and efficacy, rather than convergence on a single position (Postmes et al., 2001). In doing so, participants balanced the desire for certainty with openness to diverse perspectives, signaling that conclusions drawn from conspiratorial information may not be certain (see also Wang and Tanes-Ehle, 2023). At the same time, uncertainty amplified participants' desire for autonomy (“…which is why I think it should be up to individuals to decide”), framing vaccination as a matter of personal choice without necessarily adopting conspiratorial reasoning as an explanation or justification.

Conspiracy theories present explanations of historical events (Keeley, 1999) that are, by their nature, contrary to the publicly accepted explanation (Douglas and Sutton, 2023) yet we know little about how people parse the two privately or through interaction with others. The current research aimed to explore how conspiracy theories are deliberated on, and to investigate what this process looks like when deliberation occurs individually (privately) vs. socially (publicly). We have suggested that the question of how people rationalize conspiracy theories is potentially informed by the very properties that define them (Douglas and Sutton, 2023). If this is the case, then the constituent elements identified by (Douglas and Sutton 2023) should be evident amongst the reflections of those who interact with them in a real-world setting. Qualitative methods are often underutilized in belief research in general (Van Den Haak and Grusauskaite, 2025) and conspiracy research specifically (Nera et al., 2020), though they offer rich and nuanced insight. From participants' writing, we sought to infer the underlying processes through which people interpret certain claims as conspiratorial and examine how these rationalizations are used, both publicly and privately, to justify belief or disbelief. We considered these processes in the context of pandemic and vaccination related conspiracy theories during a period of global crisis where uncertainty was heightened. While the underlying deliberative processes may extend to other conspiracy contexts, replication across different topics is necessary to determine whether similar patterns emerge.

Conspiracy theories as the sum of their parts

Our analysis showed that people generally evaluated conspiracies holistically. That is, their essential features and how they came together to form the theory as a whole were interrogated, rather than any single element being evaluated in isolation. We observed that, for the alternative account to be construed as conspiratorial (per Study 1), participants reliably drew on three broad elements: a) the claim was interpreted as “risky” due to its extremity and perceived absurdity (plausibility); b) it highlights a power imbalance between untrustworthy “elites” and the unwitting public (trust); and c) it must describe harm inflicted on the public (harm). In Study 2, when one or more of these broader thematic components were negated during deliberation, participants were less likely to construe the claims as conspiratorial. For example, plausibility was not central to participant's deliberations (both private and public), given that they acknowledged that certain aspects of the claim were, in fact, true. With this “riskiness” element weakened, the claim no longer fit with the core criteria identified in Study 1. We therefore offer insights regarding the communication of conspiracy theories, suggesting that what “sounds like a conspiracy” to laypeople varies depending on combinations and variations of consistent structural features (Douglas and Sutton, 2023; Harmon-Jones et al., 2024; Raab et al., 2013).

The results also demonstrated that, whether conspiracy theories about the origins and spread of COVID-19 (Study 1) or the safety and efficacy of vaccination (Study 2) were discussed, different interpretations of these same structural features of the theories ultimately informed judgments for (dis)belief. While our sample was comprised predominantly of people who rejected the conspiracies, believers and disbelievers alike recognized that the conspiratorial account of events was opposed to the commonly accepted one. Similarly, they recognized that the agents implicated (whether individuals like Bill Gates and Dr. Anthony Fauci or institutions like pharmaceutical companies) must have the capability of orchestrating and carrying out the plot, and they must do so with malevolent or nefarious intentions to cause harm. Furthermore, when people believed in the conspiracies, their verbal behavior indicated that these features were considered true representations of the world. For people to reject the conspiracies, one or more of these elements was explicitly negated. This is consistent with prior research which has found that discussions of powerlessness, deception, defiance and distrust were significantly higher in online conspiracy groups compared with matched control groups (Klein et al., 2019) and scientific groups (Fong et al., 2021). That is, a potential linguistic marker of people who endorse conspiracies is that they enlist their own experiences with hierarchies and past abuses of power as evidence that there are “hidden enemies among us” (Klein et al., 2019, p. 12).

Furthermore, the role of the alleged conspirator was pivotal in shaping participants' judgments, causing perceptions of what is a “plausible” explanation to differ from context to context. In Study 1, the conspiratorial account was generally considered more implausible than the widely accepted account, because the assumption that multiple (supposedly) disconnected individuals who comprise the “wealthy elites” were able to coordinate in secret, and all with the same nefarious goals was considered extreme. While the extremity of the claims meant that the theory was construed as more conspiracy-like (see also Harmon-Jones et al., 2024; Raab et al., 2013), it also meant that the claims were readily delegitimized. In Study 2, on the other hand, many participants did not necessarily consider the conspiratorial and widely accepted interpretations to be in direct conflict. Rather, they identified elements of both accounts that seemed observationally true, such as occasional adverse effects. However, participants indicated that the conspiratorial claims were potentially plausible as they referred to an institution (pharmaceuticals) that exist within a flawed system (capitalism). In this way, they were able to circumvent the systemic problems they identified with systemic solutions. These findings highlight that rationality, as it relates to conspiracism, is “deeply contextual” (Forberg, 2023, p. 662).

The “conspiracy” label as a discursive tool

Furthermore, we found that group deliberations neither inherently encourage nor discourage conspiracy beliefs; rather, they are a receptacle in which emergent norms shape the group's approach to collective sense-making. In particular, we observed that explicitly labeling a claim as a “conspiracy theory” significantly altered the group dynamics and the willingness to engage in meaningful, nuanced discussions. In Study 1, where participants readily identified the alternative account of the origins/spread of COVID-19 as a conspiracy theory, group discussions were characterized by ridicule and disparagement, which introduced a tendency to distance oneself from any position that might be perceived as conspiratorial. Existing work on the stigmatization of conspiracy beliefs (Lantian et al., 2018; Nera et al., 2020) aligns with these findings, suggesting that the label can be employed as a rhetorical tool to delegitimize novel claims. As a result, an anti-conspiracy identity was fostered within these group chats, which might suggest effects of “identity-truth malleability”, wherein the salience of one's identity leads to perceptions of the truth that are aligned with those identities (Wang et al., 2023). That is, rather than encouraging norms for critically engaging with and analyzing the quality of the claims themselves (per the particularist view, Dentith, 2016; see also Vermeulen, 2025), the “conspiracy” label was used as a proxy for distinguishing truth from falsity.

Another possible interpretation, however, is that readily labeling the alternative account as conspiratorial within group chats may reflect a pre-existing consensus to reject unconventional viewpoints, rather than an effort to actively stigmatize other people and/or their beliefs. That is, some groups may have been comprised entirely of people who held anti-conspiracy views prior to their engagement with the group deliberation task. In this case, participants who disparage “conspiracy theorists” may not do so to persuade or influence others by stigmatizing alternative viewpoints and suppressing potential dissent. Rather, they may do so in an effort to further solidify the groups' consensus and establish a stronger sense of “us” (rational, anti-conspiracists) vs. “them” (irrational conspiracy theorists) (Ferguson and Ford, 2008).

On the other hand, we also find descriptive evidence that group discussions which foster open dialogues allow people to express healthy skepticism whilst supporting their tolerance of ambiguity, per Study 2. In the absence of the “conspiracy” label, new higher-order themes relating to uncertainty and autonomy emerged, particularly during group deliberations. Participants in these group interactions generally treated each piece of information as uniquely valuable and worthy of scrutiny, fostering dialogues that were open to nuanced perspectives rather than encouraging users to be absolute in their thinking (Newman et al., 2022). Evidence toward the usefulness of such open dialogue is potentially observed in recent work on the role of Large Language Models in correcting conspiracy beliefs and misinformation. Brief conversations with chatbots have been successful in steering people away from false beliefs, potentially because they encourage critical thinking (Costello et al., 2024), present relevant information in a non-confrontational manner (Boissin et al., 2025), and tailor responses to specific concerns, fostering agency and trust among users (Pearson et al., 2026). Indeed, creating spaces to engage in transparent discussions about controversial topics emphasizing the value of norms for independent, critical thought (Postmes et al., 2001) provides people with the opportunity to take an active (rather than passive) role in their decision-making (Lewin, 1947). Such group-based problem solving is beneficial for satisfying the need for autonomy (Ryan and Deci, 2000), which, as we have outlined above, is integral to participants' judgments about conspiracies. Taken together, these findings might suggest that, under the right conditions, small group interactions could aid in curbing conspiracy theorizing.

Limitations and future directions

We have adopted a novel approach of inviting real-time interactions about conspiracies and analyzed the naturalistic data using qualitative methods (which are underutilized in this area; see Nera et al., 2020). Nevertheless, our approach is not without limitations. First, less than 14% of our total sample agreed with the COVID-19 and vaccination conspiracies. As such, our findings disproportionately represent the perspectives of those who disagreed – an imbalance that potentially reflects the fringe nature of these beliefs. The skewed distribution of opinions is undesired in that it restricts our ability to draw conclusions about how conspiracies are socially negotiated, particularly from believers' perspectives. However, we note here that increasing concerns in contemporary social psychology to normalize data so that they fit statistical and methodological expectations is potentially counterintuitive to the progression of research (Haslam and McGarty, 2001). That is, some social phenomena are complicated and unbalanced in the real world. Artificially adjusting the distribution, then, may reduce statistical uncertainty, however, it may also “exceed the capacity for the [phenomena] undergoing investigation to be normalized” (Haslam and McGarty, 2001, p. 17). Indeed, US polling conducted during the collection period of the current data observed that between 10 and 25% of people believed that Bill Gates was behind the pandemic or that coronavirus was used to force a dangerous vaccine on Americans (Uscinski et al., 2022). Future experimental conspiracy research (and particularly that which is concerned with the social communication of conspiracy theories) must therefore contend with the struggle between providing acceptable estimates of statistical uncertainty as well as a realistic depiction of how conspiracy beliefs are represented in the world.

Second, to avoid priming participants, we decided to post-measure conspiracy endorsement. Prior beliefs are therefore a rich source of random variation in the data. Furthermore, as our design involved the use of live, unmoderated group chats, we lacked control over the characteristics of participants within the groups. The effects of different configurations of characteristics (for example, groups that unanimously agree, groups that unanimously disagree, groups that contain multiple viewpoints, etc.) should be confirmed in future studies. Furthermore, the timing of the tasks (7 vs. 10 min) resulted in a discrepancy in the amount of text produced across conditions. That is, participants who deliberated alone produced less text on average than the cumulative text of those who deliberated in groups. Although this difference was introduced to accommodate group discussion while minimizing fatigue for those in the individual condition, it may have contributed to effects by increasing elaboration. Future research may consider holding deliberation constant or modeling text production rates. We note also that our data were collected via MTurk – future research should involve samples obtained from other sources to assess whether effects generalize (Coppock, 2019).

Finally, specific reference to Dr. Anthony Fauci and Bill Gates as examples of “elites” implicated in the conspiracy account of Study 1 may have biased participant responses, given that these are politically salient figures. Evidence suggests that, while mention of specific “elites” may evoke political asymmetries in endorsement, the relationship between political ideology and conspiracy belief is complex (Enders et al., 2023a). For example, Enders et al. (2023a) observed that the specific belief that COVID-19 was caused by Bill Gates was endorsed at comparable rates by Democrats, Republicans, and Independents alike, indicating that how conspiracies are disseminated (and by whom) may be a significant factor in activating political bias. Exploring how the source of conspiratorial information interacts with the implicated figures is therefore a potential avenue for future work.

Conclusion

The current research suggests that conspiracy theories are interpreted (in the literature, as well as in the “wild”, according to those who interact with them) through consistent key features. Judgements about whether to accept or reject them are, in turn, shaped by combinations and variations of these features. In keeping with prior research, we find descriptive evidence that feelings of distrust for scientific, pharmaceutical, and government institutions, as well as the people who hold positions of power within these institutions, are foundational for belief in conspiracy theories. We observe that group interactions have the potential to both help and hinder the dissemination of conspiracy theories (Uscinski, 2018), depending on the norms that emerge among group members. Attention to processes of collective sense-making is therefore instrumental (Van Den Haak and Grusauskaite, 2025). Our findings underscore how conspiracy theories are communicated, highlighting that collective judgements are shaped not only by the content of the claims, but the ways in which they are famed and negotiated during interaction. We suggest that practices of thoughtful, nuanced discourse may help to mitigate backlash effects associated with pejorative use of the “conspiracy” label, while supporting people to take a more active role in their decision-making.