Endorsement of increased-risk claims about Tylenol and autism may not reflect causal beliefs

A recent KFF poll prompted concern that many Americans believe it is definitely or probably true that Tylenol use in pregnancy causes autism. However, the poll actually asked a slightly different question: whether Tylenol use increases the risk of autism, raising questions about how such phrasing is interpreted. We conducted a preregistered experiment (N = 1957) comparing three phrasings that appeared in news coverage of the poll: increases risk, causes, and a mere link. Overall, differences in phrasing had minimal impact on endorsement. Instead, responses varied substantially by respondent characteristics, particularly political party, replicating the poll's finding that Republicans were more likely than Democrats to endorse the relationship. The only evidence of a phrasing effect was conditional on education: among more educated respondents, endorsement was lower when the relationship was described as explicitly causal, but similar for the increases-risk and mere-link phrasings. These results suggest that caution is warranted when interpreting the public's causal beliefs, and further work is needed to develop measures that more clearly distinguish beliefs about association from beliefs about causation.