Correlation Does Not Imply Causation. What Does? (rus)
Date:
An introductory talk on causal inference. Our thinking is poorly adapted for reasoning about cause-and-effect relationships. This leads to problems in work communication and introduces distortions in the modeling process.
For example, it’s easy to notice the connection between health and frequency of hospital visits - those who often visit doctors are sick longer and more severely. In most cases, however, doctors improve rather than worsen health - which contradicts the above observation.
The talk discusses why it’s so difficult to think about conditional probabilities and how to deal with it.
Original title: “Корреляция не подразумевает причинно-следственную связь. А что подразумевает?”