correlation != causation

Tags: probability

• https://www.gwern.net/Correlation
• https://en.wikipedia.org/wiki/Berkson%27s_paradox
• https://www.gwern.net/Everything

• At large scales, everything is correlated
• Interrcorrelation implies causal networks
  • A correlates to B, B correlates to A, or A correlates to C + C correlates to B
    • This is most often wrong, since it’s actually:
      • A correlates to B
      • B correlates to A
      • A correlates to C1, B correlates to C1
      • A correlatees to C2, B correlates to C2, etc
  • Relationships are directly casual, or they can flow through a huge number of causal chains