A variable that influences both the independent and the dependent variable, distorting the association between them: it can create a spurious link, hide a real effect or even reverse it.
Card 1 of 259. Answer: A variable that influences both the independent and the dependent variable, distorting the association between them: it can create a spurious link, hide a real effect or even reverse it.
About · Quizzes · History of AI · To suggest new questions or report an error send me a dm.
A confounding variable affects both the independent and the dependent variable, so their observed association is not purely the effect of one on the other. It can create a link where none exists, hide a real effect or reverse it. Ice cream sales and drownings rise together because hot weather drives both.
A 1986 kidney stone study shows how strong this can be. Open surgery succeeded in 78% of 350 patients and percutaneous nephrolithotomy in 83% of 350, yet surgery did better for both small stones (93% vs 87%) and large stones (73% vs 69%). Surgery went mostly to large, harder stones, so stone size drove both treatment and outcome. The reversal is Simpson's paradox.
Randomization breaks the link between confounder and treatment, which is why A/B tests are the gold standard. With observational data you can stratify, match or adjust for measured confounders, but unmeasured ones remain. Adjusting for the wrong variable also backfires: a mediator, on the causal path, or a collider, a common effect of both.
