Introduction

Sometimes labelling something as a “conspiracy theory” is intended as a way to shut down debate; it becomes part of a set of techniques used to neutralize, minimize, or attack a claim. For example: former UK Prime Minister Tony Blair and US President George W. Bush cast criticism of the invasion of Iraq in 2003 as belief in a conspiracy theory (Saraceni, 2003). A defender of Lance Armstrong against doping charges, Rep. Jim Sensenbrenner of Wisconsin, dismissed accusations against the cyclist as “conspiracy theories” (Vertuno, 2012). And both sides of the political aisle claimed the various theories either alleging or denying that foreign interference in the 2016 US Presidential election were just “conspiracy theories” (Uscinski and Enders, 2022).

Yet we might also think these very examples speak against the thesis that labelling something as a “conspiracy theory” stops or puts a halt to debate: despite the protestations of the UK and US governments in 2003, it was eventually accepted that elements within those governments tried to cover up a lie justifying the invasion of Iraq (that the Saddam Hussein regime had continued development of Weapons of Mass Destruction). After repeated, vehement denials, the evidence and testimony accumulated, and Lance Armstrong finally admitted that he was guilty as charged. Finally, whatever we think about the 2016 US presidential election, the fact that people labelled the various competing views as “conspiracy theories” certainly did not stop investigations into whether those claims were true.

These few examples suggest that the label “conspiracy theory” is complicated. It can alter people’s beliefs in claims, and has the rhetorical power to shape or silence political discourse both formal and informal. Thus, empirical research is required in order to understand what, if anything, results from calling something a “conspiracy theory.” Yet, lately, researchers across many disciplines have claimed that the phrase has no effect, and they cite in passing Michael Wood’s 2016 article “Some Dare Call It Conspiracy: Labeling Something a Conspiracy Theory Does Not Reduce Belief in It.” His paper argues that the label “conspiracy theory” does little or no discursive work, and does not influence people’s beliefs.

This paper reconsiders Wood’s findings. First, we examine the contribution and salience of Wood’s work. Second, we present the results of our own survey-cum-experiment to reopen the case: the phrases “conspiracy theory” and “conspiracy theorist” appear to do some work to influence belief after all. We argue that researchers need a more refined understanding of the varied and complex social, political, discursive, and interactional effects of the phrase.

Dismissing the Power of a Label

When we consider the label “conspiracy theory” we sometimes ask “Do people believe these things called ‘conspiracy theories’?” whilst at other times we ask “Does the label ‘conspiracy theory’ affect whether people are willing to take up or admit to belief in conspiracy theories?”

The first question is about believability: if something is labeled a “conspiracy theory,” does that make people more or less inclined to believe it? If I label the view that “Lee Harvey Oswald did not act alone in killing President Kennedy” as a “conspiracy theory,” does this affect whether people will adopt that view as a belief?

The second question is about rhetoric: does the label “conspiracy theory” affect whether people are willing to admit to belief in, or even consideration of, a given conspiracy theory? That is, if other people call the view that “Lee Harvey Oswald did not act alone in killing President Kennedy” a “conspiracy theory,” does this make me more or less inclined to admit that I believe it?

Wood, in his 2016 paper, created two surveys to test whether the label “conspiracy theory” changed respondents’ belief in a conspiracy claim, and found no evidence that it did; he was, in effect, considering believability. That is, the label — according to Wood — has no interactional or discursive power.

Wood, however, was also challenging work by researchers who were motivated to investigate the rhetorical power of the label. As he argued at the time, researchers were claiming that:

Calling something a conspiracy theory (or someone a conspiracy theorist) is seen as an act of rhetorical violence, a way of dismissing reasonable suspicion as irrational paranoia. For deHaven-Smith (2013), the conspiracy-theory label comes with such negative baggage that applying it has “the effect of dismissing conspiratorial suspicions out of hand with no discussion whatsoever” (p. 84). Husting and Orr (2007) likewise argued that applying the label “discredits specific explanations for social and historical events, regardless of the quality or quantity of evidence” (p. 131). (2016, p. 695).

Wood’s work was, then, motivated as a means of further testing a claim which was about rhetoric and not believability that other researchers had previously investigated, both theoretically and, crucially, empirically.Footnote 1 This is, itself, a problem, as Wood shifts the goalposts from talk of the rhetorical power of the label to its role in diminishing belief. However, the bigger issue is that, increasingly, a small but rising number of articles have overgeneralized the strength and scope of Wood’s findings about how the label “conspiracy theory” affects believability without critical discussion of that prior work concerning its rhetoric power, as though his data (as opposed to the analyses which came before it) is the final answer to the question concerning the power of the phrase “conspiracy theory.”

Examples that treat Wood’s conclusion as a fait accompli emerge across disciplines, particularly in psychology. Jan-Willem van Prooijen and Mark van Vugt claim: “Although in modern times conspiracy theories can carry a social stigma (Harambam and Aupers, 2015), using the label ‘conspiracy theory’ does not decrease people’s belief [emphasis ours] in it (Wood, 2016)” (van Prooijen and van Vugt, 2018, p. 775). Similarly, Biddlestone et al. (2021) write “evidence suggests that labelling an idea a ‘conspiracy theory’ does not affect its believability [emphasis ours] (Wood, 2016)” (Biddlestone et al., 2021, p. 8). Biddlestone et al. are correct that Wood’s evidence suggests this; what they and other researchers do not acknowledge is that limited experiments do not make an airtight case. Referencing Wood, Cíbik and Hardoš claim that the label might have lost once-pejorative effects: “It seems this delegitimizing strategy of labelling something a conspiracy theory may no longer reduce belief [emphasis ours] in it” (Cíbik and Hardoš, 2020, fn. 3). Despite the qualifiers “suggests,” “seems,” and “may,” passing references such as these allow for a premature closure of inquiry. They seem to make an open case appear closed; as though there is little problem or power behind the label “conspiracy theory.”

Such citations of Wood seem to be spreading into adjacent literatures like religious studies, and library and information science, and we see kindred citations in the philosophical literature on conspiracy theory. Phillipe Huneman and Marion Vorms, for example, write: “But see Wood 2016, who shows that nowadays characterizing a view as a CT is, on the average, not detrimental to its acceptance and diffusion (2018, fn. 3).”Footnote 2

Indeed, much of the debate that cites Wood in the philosophical literature centers on how to properly define what counts as a “conspiracy theory,” and thus how to properly understand whether belief in such theories is prima facie irrational.Footnote 3 This debate has rested on a distinction between generalism (the view that “conspiracy theories” can be rejected a priori) and particularism (the view that conspiracy theories are not inherently irrational beliefs, and thus must be judged on particular merits or faults).Footnote 4 Wood’s findings have been seen as evidence that people do not reason like generalists; ordinary reasoners do not treat what is labeled as a “conspiracy theory” as marking out an irrational belief (e.g. see Basham’s 2016). Wood’s work has become an iconic citation in scholarship on conspiracy theorizing; so much so that Pierre (2020) used it as support for the claim that the phrase does have pejorative power or can enhance believers’ commitment to conspiracy claims. But what does Wood’s work actually show?

Wood’s Survey Experiments

Wood used two extremely focused survey experiments to test whether the phrase “conspiracy theory” could influence believability (whether labelling a claim a “conspiracy theory” would change the likelihood that respondents would believe said claim). The first of the two experiments traced 150 participants’ responses to news-like stories of actual and fabricated conspiracy claims. He split the sample roughly in half; 67 respondents read accounts using the phrase “conspiracy theory” whilst the other half read accounts using an alternate descriptor (labelled as “ideas”). Wood found no statistically significant evidence that labelling one story a “conspiracy theory” over the other changed anonymous respondents’ likelihood to believe that the story was true.

Wood’s second experiment asked a more robust sample of 800 US participants to evaluate the believability of claims from a mock news article on a fictional Canadian political scandal. In the experimental condition, the scandal was described once, midway through the article, as a “conspiracy theory.” For the control group, the scandal was labeled “corruption allegations.” Wood found that the change in wording once again provided no statistically significant effect on US respondents’ opinions about the ersatz conspiracy. These results, alongside the first experiment, led Wood (and many who cite his work) to “suggest that the conspiracy-theory label possesses far less rhetorical power than previously assumed” (2016, p. 702). While Wood’s method and data were concise, two limitations of his study counsel against blanket generalizability of his findings, and should invite researchers to conduct more refined empirical research and more careful theoretical development.

The first limitation concerns the conflation of two distinct phenomena. What Wood measured (in a limited way) was “believability”; his two survey experiments tested whether or not the phrase can dissuade respondents in surveys from anonymously agreeing with particular claims. But Wood, and those who cite him, often conflate the power to influence someone’s belief with other forms of influence or power of the phrase. Measuring how words change people’s belief in a claim — “believability” — is not the same as measuring the varied kinds of rhetorical power it can have.

For example, people can use the term to try to lower the standing of someone else regardless of whether it affects anyone’s level of belief in a claim. What’s more, use of the term may succeed regardless of whether any minds are changed. The power to influence belief is not equivalent to discursive or rhetorical power. Conflating believability with rhetorical power, or the term’s potential to silence dissent, misses a large part of what previous researchers have flagged as a set of epistemological and political problems in relation to the phrases. Yet Wood claimed in 2016:

Rather surprisingly, there has been no empirical investigation of whether the conspiracy-theory label actually has the impact people assume it does: No one has actually investigated whether calling something a conspiracy theory makes people believe it less (2016, p. 696).

Wood is correct here; but his scope or frame with respect to believability alone is narrow.

After all, this distinction between believability and rhetorical power is important if we are interested in the effect the label “conspiracy theory” has on public debate (rather than, say, how it affects someone in a test situation). Wood argues:

Two experiments found no evidence of a negative effect of calling something a conspiracy theory. Experiment 1 showed no evidence that the label had any effect on endorsement of general conspiracist views or beliefs in real historical conspiracies, and Experiment 2 failed to find an effect with a fictitious political scandal previously unknown to participants and a large enough sample to detect a small effect with 80% power (2016, p. 702).

Labelling something a conspiracy theory may or may not affect its believability, but the way in which the label is deployed as part of a rhetorical strategy in a public (rather than test) setting is a different affair, given such labelling practices in the social world are often enmeshed in notions of power, especially notions of who has, and hasn’t, the power to deploy labels effectively.Footnote 5

The second limitation of Wood’s experiment concerns the complex set of venues, contexts, and goals that can accompany the phrase. A survey can precisely target variations in phrasing, but it also strips out the kinds of contextual cues and circumstances that are at the very heart of how the phrase can have rhetorical power to shape belief (or to shape public response independent of belief) to a claim of conspiracy. We point here not to any failing of Wood’s work, but to the seemingly rote citations of it subsequently as definitive confirmation that the phrase has no pejorative effects. The experiments thus described are far too narrow to have tested the overall discursive power of the term tout court, even just in terms of believability. Context matters.

In a sense, this is the problem of the lab vs. the field: the phrase can be used in complex ways that outstrip Wood’s measurement. This seems to be an instance of the classic problem of ecological validity; the results we get in two tightly controlled experimental conditions are not necessarily generalizable to phenomena we find in the field. After all, we should ask: Where, how, in what tone of voice, and by whom is the phrase used? What is the context of the utterance? An informal conversation differs from a meeting of Parliament or a news media story about a fake conspiracy. Who are the participants and witnesses? Who is speaking to whom, and who is listening, assumed to be listening, or assumed to be out of earshot? What are the motives and intentions of the speaker (as previously noted)?

So, with all that in mind, what exactly does Wood’s work show? In fact, it supports the conclusion that in Wood’s very specific circumstances the label has little or no power to affect believability. He used one version of the phrase as a (weak) prompt: a single occurrence of a phrase in a news article about a Canadian political event in which the US audience had little or no investment. He did not test for variations in frequency, context, or who used the phrase about whom. Tone and medium count: print, political debate, casual conversation, a tweet. All these variations matter if we are testing what, exactly, the label “conspiracy theory” does in discursive or rhetorical context. To be clear: Wood’s survey design, and his framing of the rhetorical power of the phrase as an empirical question about how the invocation of the phrase affects believability is an important contribution to the empirical work on conspiracy theory. However, the conclusions from his two surveys neither effectively settle the question of believability in relation to the phrase generally, nor do they measure the manifold ways the phrase is used to try to assert political or rhetorical power, or to silence dissent. Thus, the rising number of citations of Wood as definite support of that conclusion is troubling, especially in the face of other theoretical and empirical research both before and after Wood’s article that indicates the phrase has considerable power.Footnote 6

Methods and Findings

We want to underline the value and significance of what Wood investigated; we come not to bury Wood’s work but to engage with it. Whether the phrase could change someone’s anonymous opinion about general and specific conspiracy claims is, after all, an important question (one we will return to at the end of this article). But one study does not definitive results make. In response, we adopted one of Wood’s key questions: Does labelling a claim a “conspiracy theory” make people less likely to believe (or say they believe) it is true? We, like Wood, turned to a survey with a quasi-experimental design.

We surveyed 1000 respondents on conspiracy theories, using a sample of adults in the USA drawn by Qualtrics. Respondents were offered modest remuneration of roughly three dollars US. We established quotas for gender and ethnicity, to ensure that on these variables we would have a representative sample. Results of demographic questions also track US Census data (with the usual exception of income, as those with exceptionally high incomes are far less likely to be included in sampling frames, and less motivated to complete a survey for very little compensation). We also scrubbed incomplete and inattentive surveys (respondents who finished in less than half of the median completion time).

The independent variable, whether a claim is labelled a “conspiracy theory” or not, was operationalized by using three versions of a question probing people’s opinions about the assassination of John F. Kennedy. We have chosen the JFK assassination conspiracy theory since it is widely known and iconic in the USA; it has enduring majority support among the public, and our assumption was that people’s beliefs would be long-standing and less malleable — less likely to be influenced by the way the question was framedFootnote 7. Each respondent was randomly assigned one of three statements.

The first version of the question, as a control, avoids the phrase “conspiracy theory.” It simply offers the statement: “The assassination of John F. Kennedy involved individuals other than or in addition to Lee Harvey Oswald.”

The second version of the question does refer to the belief as a “conspiracy theory”: “The ‘conspiracy theory’ is justified – John F. Kennedy’s assassination involved individuals other than or in addition to Lee Harvey Oswald.” Here we introduce the phrase, but the scare quotes clearly imply that those calling it a “conspiracy theory” are using the term as a pejorative, are wrong in doing so, and the respondent is invited to share the perspective.

The third statement also uses the phrase as a pejorative, to discredit the claim in a more intentional and more dismissive fashion: “The unjustified belief that John F. Kennedy’s assassination involved individuals other than or in addition to Lee Harvey Oswald is just another conspiracy theory.” The wording invites respondents to be “one of us,” part of an in-group, or one of “them,” part of an out-group. (This version was reverse-coded, since “true” indicates strong support for the “lone gunman” theory that Oswald acted alone.)

Respondents were randomly assigned one of these statements: 335 respondents were presented with the first version, 327 with the second, and 338 with the third. We will abbreviate this variable, the JFK Question asked, as JFKQ.

Once presented with their particular statement, respondents used an 11-point slider scale to indicate that they were absolutely confident the statement was false (“0”), unsure (“5”) or absolutely confident that the statement was true (“10”). This response is the dependent variable “Support for the JFK Conspiracy Theory” (JFKCT) (Table 1).

Table 1 JFK conspiracy statements
Full size table

To offer an effective descriptive, we can assemble a cross-tabulation, collapsing the 11-point scale used to measure JFKCT into three groups: strong, moderate, and weak support for belief that Oswald was either a patsy, or did not act alone. For these categories, 0–2 was coded as weak support, 3–7 as middling, and 8–10 as strong support.

Of those asked the first version of the question (with no mention of conspiracy theories), 15.5% expressed weak support for the belief that there were co-conspirators, 36.7% indicated moderate support, and 47.8% indicated strong support.

For those who were asked the second version of the statement, where “conspiracy theory” is acknowledged as a pejorative, but unfairly applied to the belief in the conspiracy, there was little change: 16.5% expressed weak support for the belief that there were co-conspirators, 32.1% indicated middling support, and those indicating strong support actually increased to 51.4%. By enclosing the phrase in scare quotes, “conspiracy theory” is not used to denigrate the claim. By recognizing how the phrase is used, that context seems to inoculate against the pejorative. Here, introducing the term has no significant impact.

For those who were asked the third version, which uses “conspiracy theory” as a pejorative, and explicitly dismisses the claim as “just a conspiracy theory,” there was a marked change. Now, those expressing weak support for the belief that there were co-conspirators more than doubled, from 15.5 to 37.0%. Those indicating strong support fell by about 60%, from 47.8 to 29.6% (χ2 = 65.1, p < .000) (Table 2).

Table 2 Support for the JFKCT by question asked
Full size table

The 11-point scale was designed to allow us to run a one-way ANOVA to determine if there was statistically significant variation in the mean of the responses to the three statements. Here, we find the mean of the responses to the first two questions were likewise quite similar. For the first, the mean response was 6.25 (SD 2.86), and for the second, it was 6.22 (SD 2.93). The third response, to the statement that rejection of the “lone gunman” explanation is “just a conspiracy theory,” was significantly different. The mean for this group was 4.78 (SD 3.10). While simply labelling something a conspiracy theory has no demonstrable effect (which is consistent with Wood’s findings), dismissing it as “just” a conspiracy theory changes things rather dramatically — at that point people are far less likely to endorse the conspiracy theory. This pattern holds across demographic categories (Tables 3 and 4).

Table 3 ANOVA: JFK response
Full size table
Table 4 Comparison of means
Full size table

We must be cautious in assessing these results. What we know for sure is that, on average, respondents moved a slider-scale differently on the third statement relative to the first two. Exactly why people’s behavior in response to the third statement tends to differ is difficult to demonstrate with certainty, and our design can only contribute in a small way to addressing that important question. We can’t determine from this what people mean by the term “conspiracy theory,” how they might use the term, or what they “really” believe with regard to the JFK assassination (or if there are differences in their willingness to admit to their belief, even if anonymously and only to the researchers). One might imagine that there are other differences beyond whether “conspiracy theory” is included and the way it is used that might affect how people respond to one or another of these statements. Although this was the last question in a long survey, and few are expected to have taken the time to parse their assigned statement thoughtfully, perhaps the tension or ambiguity between what we think count as “theories” and “beliefs” contributes to way people respond to these (see Duetz (2022) for a discussion of this concern). Nonetheless, we do think we have some window into whether people and how people react to the term. Despite that this is taking place in an artificial setting, and although a quantitative approach risks sacrificing nuance, clearly something is going on.

Discussion and Conclusions

Wood has opened an interesting and important line of inquiry, and does us a service by interrogating whether the claim “conspiracy theory” is necessarily a pejorative term. If it turns out the phrase is commonly used as an epithet which seeks to serve to silence debate, then we should continue the train of thought and ask whether the epithet is actually effective in changing people’s hearts and minds. Regardless of the intent, it is important to inquire what the effect of being called a “conspiracy theorist” has, if any, on one’s beliefs.

Our research, however, did not support Wood’s conclusion that calling something a “conspiracy theory” has no effect upon people’s belief in a claim. We find that introducing the label can indeed affect the believability of even a presumably well-established opinion.

It is important to note, though, that our study, like Wood’s, has its limitations. While we have tried to introduce context in our phrasing of our prompts, this remains an experimental design. Whether a researcher’s question will affect behavior (especially a response to a Likert scale question) in the same way that the invocation of “conspiracy theorist” does in political discourse or casual conversation is unlikely.

As such, neither Wood’s research, nor ours, should be taken as the last word on the topic of the power and scope of the label “conspiracy theory” when it comes to public debate or the academic debate between generalist or particularist takes on the warrant of conspiracy theories. There is still more work to be done (as we suspect Wood would be the first to agree). More, empirical work, using multiple methods, will help clarify how people perceive the phrases “conspiracy theorist” and “conspiracy theory,” and whether their use influences belief. Further investigation into what other kinds of power are in fact (not just in theory) at stake when people use the terms in specific contexts could also be the subject of continuing study. In an anonymous survey, people may not be swayed much by what they (may or may not) perceive as an attempt to belittle (or reinforce) a position they hold. In face-to-face interviews, however, or ethnographies (even if confidentiality is assured), a respondent may be more careful to avoid appearing to be one of “those” people.

What we can say is that, along with scholars preceding and following Wood, we support the claim that “conspiracy theory,” at least in some cases, performs discursive work; saying that something is “just a conspiracy theory” affects the way people respond in a range of contexts. That is, contrary to Wood, our work suggests that some people do reason like generalists. The nature, strength, and variability of that effect — the question of whether deploying the phrase “only” shuts down discussion, or whether it also dissuades the “conspiracy theorists” from believing in them — is not, however, entirely clear.

None of this negates Wood’s findings, but it should make us more cautious of treating Wood’s work, or anyone’s, as a final word, as proof conclusive. Instead, his work is provocative, suggestive, and an important contribution to an interesting line of research. However, declaring that the phrase “conspiracy theory” has no measurable rhetorical effect — as some particularists and many generalists have done — is premature. It is, at this stage, uncertain exactly what the label is doing. The contexts under which it is uttered, and the kind of effects the accusation can have on discourse are multiple and various. So, until such time as we have more fulsome results which are less ambiguous, we should be cautious in our citation of Wood. Despite findings which are far less than conclusive, we do hope that, as a compliment to Wood’s analysis, our new research serves to caution reliance on a limited number of studies, and spur continuing study of the supposed impact of the epithet.

Still, with all that said, here is a quick way we can all contribute to this research (especially if one is unsure that the invocation of the accusation has no effect): the next time an opportunity arises, in an elevator, a subway, a business meeting or a cocktail party, note what happens when you call someone a “conspiracy theorist.”