Scott Alexander warns about the potential misinterpretation of odds ratios in studies, explaining how to convert them to effect sizes for more accurate understanding.
Longer summary
Scott Alexander discusses the potential for misinterpreting odds ratios in statistical studies, using a personal anecdote from a journal club. He explains how odds ratios can seem more significant than they actually are, and provides a method for converting them to effect sizes for better interpretation. The post includes a reference to Chen's study on interpreting odds ratios in epidemiological studies and gives an example of how a seemingly impressive odds ratio can translate to a more modest effect size. Scott emphasizes the importance of careful comparison between studies that report results using different metrics.
Shorter summary