Never Tell Me The Odds (Ratio)

A short statistics-literacy note (low-confidence epistemic status) on a real numeracy trap: odds ratios sound bigger than they are. Scott almost over-rated an antidepressant study reporting an OR of 2.9 until catching himself, and shares Chen's method for converting odds ratios to effect sizes (a quick heuristic: ln(OR)/1.81). The worked example is clean: a trial where the drug doubles the recovery rate (300/1000 -> 600/1000) yields a relative risk of 2.0, an odds ratio of 3.5, and an effect size of only 0.7 -- still not a 'large' effect by convention. The moral: odds ratios inflate the apparent magnitude and effect sizes deflate it, so be careful comparing studies that report results in different units. A useful, memorable practical heuristic, but very short and conveying an existing formula.

Quality 60 · Strong. Low-Strong (60): a genuinely useful and memorable stats-literacy point (don't be fooled by odds ratios; the OR->effect-size conversion) with a clear worked example -- but very short (~330 words) and conveying an existing formula rather than developing an idea, which keeps it at the Solid/Strong boundary.

Claude’s paradigm shift 35 · Slight. Slight (35): the practical OR->effect-size translation is a useful flag, but it conveys Chen's existing method rather than offering anything novel.

Real-world impact 2 · Minor. A useful, memorable stats-literacy point (don't be fooled by odds ratios; the OR-to-effect-size conversion); a short conveyance of an existing formula within the discourse, no material reach → RWI 2.