Hacker News new | ask | show | jobs
by ovi256 2842 days ago
Before conducting controlled experiments (which can verify causality) it's usual to conduct compared studies, which find the experimental factor occuring in a population and study its outcome. It's enough to establish correlation, much cheaper, and you can do it today instead of waiting however long it takes for the experiment (years if you have to grow trees while exposing them to microwave radiation)
1 comments

True, but when you find a lot of random change correlations that are not really there.