Because an association can come from a confounder that drives both variables, from causation running the other way, from how the data were selected, or from chance. Only an experiment, or careful causal analysis of observational data, shows that changing one changes the other.
Card 1 of 244. Answer: Because an association can come from a confounder that drives both variables, from causation running the other way, from how the data were selected, or from chance. Only an experiment, or careful causal analysis of observational data, shows that changing one changes the other.
About · Quizzes · History of AI · To suggest new questions or report an error send me a dm.
A correlation says two variables move together, not why. Four explanations compete with "X causes Y": a confounder Z causes both; Y causes X (reverse causation); the sample was selected in a way that links them; or the pattern is chance, which becomes likely when you test many pairs.
Example: countries that eat more chocolate have won more Nobel prizes per capita, with a correlation of about 0.79 across 23 countries (Messerli, New England Journal of Medicine, 2012). Chocolate doesn't make laureates; national wealth is a plausible confounder that raises both chocolate consumption and research funding.
Randomized experiments, such as A/B tests, settle causation because assignment can't depend on anything else. With observational data, causal inference methods such as matching, difference-in-differences, instrumental variables and regression discontinuity get closer under stated assumptions. A good follow-up: smoking was shown to cause lung cancer without a randomized trial, from the strength, consistency and dose-response of the evidence that Austin Bradford Hill summarized in 1965.
