Statistical misreasoning in online content about vaccines: Implications and recommendations for addressing disinformation.

Abstract

Statistical misreasoning is a key mechanism through which anti-vaccine narratives distort scientific information and undermine public confidence in immunisation. Although prior research has examined thematic and ideological features of vaccine misinformation, little is known about the specific errors in numerical reasoning that shape users' interpretations of vaccine-related data. A total of 597 Polish-language Facebook posts expressing anti-vaccine views and containing references to statistical information were analysed. Based on previous research on statistical cognition and an inductive review of the material, a coding scheme comprising ten categories of statistical misreasoning was developed and applied to all posts. Quantitative analyses were then conducted to examine how frequently these categories occurred and which combinations of errors appeared together. The most prevalent forms of misreasoning were the correlation-causation fallacy (70%, p < 0.001) and base rate neglect (58%, p < 0.001). Denominator neglect and cherry picking appeared in half of the posts, while the remaining categories were less frequent. Most posts contained multiple errors (median = 4), and the most common configuration involved the correlation-causation fallacy, base rate neglect and denominator neglect. The distribution of error counts further showed that posts most often exhibited four distinct categories of misreasoning (23%), followed by three (19%) and five (17%), and overall a majority of posts (62%, p < 0.001) contained between one and four different types of errors. Co-occurrence analysis revealed stable structural patterns, with the strongest association observed between denominator neglect and intuitive reasoning error (ϕ = 0.23; p < 0.001). Anti-vaccine discourse exhibits systematic patterns of statistical misreasoning that shape erroneous interpretations of vaccine-related data, highlighting the need to address cognitive and statistical misunderstandings through targeted public health communication.