A recent study suggests that the phrasing of poll questions can influence how people interpret scientific claims about health risks, particularly when such issues are subject to public debate. The findings were published in Scientific Reports and announced on Jul. 5.
Researchers focused on whether different ways of wording a question about Tylenol (acetaminophen) use during pregnancy and autism risk would affect how strongly people endorsed the claim. The study did not evaluate any biological link between acetaminophen and autism, but instead examined interpretation of language such as 'is linked with,' 'increases the risk of,' and 'causes.'
The experiment involved 1,957 participants recruited from CloudResearch’s Connect platform. Participants answered questions about their awareness of Tylenol use during pregnancy and autism, then indicated their level of agreement with differently worded statements. Researchers found that endorsement rates for ‘is linked with’ and ‘increases the risk of’ were similar, while statements using ‘causes’ received less strong support. College-educated respondents showed greater skepticism overall and expressed weaker support for causal language compared to associative phrases.
The study also analyzed responses by political affiliation and educational attainment. While Republicans, Democrats, and Independents differed in their overall likelihood to endorse the claim, all groups responded similarly across different phrasings. Republicans had about 9.8 times higher odds than Democrats to endorse the statement; Independents had 3.25 times higher odds.
Researchers noted that subtle changes in survey wording may have only modest effects on endorsement strength but could still shape public understanding or misinterpretation of scientific information. They emphasized that greater educational attainment was associated with better differentiation between correlation and causation.
The authors concluded that future research should focus on developing measures to more clearly distinguish between beliefs about association versus causation among broader populations.
