Towards Caution in Generalisation: How to Draw Conclusions About Conspiracy Believers Based on Mostly Non-Believer Samples

The psychology of conspiracy theories is built on convenience samples of individuals who mostly reject conspiracy theories. Inference about strong conspiracy believers – who are largely absent from research samples – rests on two generalisability assumptions. The first, interpretation generalisability, assumes a sample’s average level of conspiracy (dis)belief does not significantly alter the theoretical interpretation of relationships. The second, range generalisability, holds that statistical relationships observed among low-to-moderate conspiracy believers are generalisable to stronger believers. Recent work on contradictory conspiracy beliefs shows that these assumptions can be misleading, highlighting the need to reflect on the conditions of generalisations. In the case of interpretation generalisability, we propose that relationships observed among low-to-moderate believers might be better interpreted as capturing differences in plausibility perceptions – because what differentiates people who “neither agree nor disagree” with a conspiracy theory from people who reject it is arguably the fact that they are open to the possibility that it could be true. Such relationships are informative, yet the differences between perceptions of plausibility and factuality (i.e., the sense that a conspiracy is true) should be considered before generalising conclusions to strong believers. As for range generalisability, we argue for a case-by-case evaluation for the possibility that relationships might be non-linear across the full range of agreement scores.