3 ms·

> correlation An interesting result, to be sure, but correlation does not imply causation.

• tomrod 7y ago
  But it may strongly hint at it and warrant further research. [0] https://xkcd.com/552/ https://xkcd.com/552/ [1] http://bayes.cs.ucla.edu/WHY/ http://bayes.cs.ucla.edu/WHY/
• rockinghigh 7y ago
  These studies only report on the evidence of a statistical relationship and not a guaranteed causal link. They also don’t use correlation but usually some sort of regression that includes other factors like temperature to remove some of the effects of these variables.
  • tomrod 7y ago
    Linear regression is a form of correlation. The model causality is implied but the only connection is correlation. Unless estimating an average treatment effect or diff-in-diff type framework, causality will be hard to identify in observational frameworks