Correcting Statistical Misinformation About Scientific Findings in the Media: Causation Versus Correlation

Although retractions significantly reduce the number of references people make to misinformation, retracted information nevertheless persists in memory, continuing to influence reasoning. One hundred and twentyninelay participants completed an adaptation on the traditional continued influence paradigm, which set outto identify whether it is possible to debunk a piece of common statistical misinformation: inappropriatecausal inference based on a correlation. We hypothesized that participants in the correction condition wouldmake fewer causal inferences (misinformation) and more correlational inferences (correction) than those inthe no-correction condition. Additional secondary hypotheses were that the number of references made tothe misinformation and correction would be moderated by the level of trust in science and scientists, and theamount of television that participants watch. Although the secondary hypotheses were not supported, thedata strongly supported the primary hypotheses. This study provides evidence for the efficacy of correctionsabout misinformation where correlational evidence has been inappropriately reported as causal