Suggestions

:bulb:

Causal inference

I use Exist to track life problems: low mood, poor sleep, etc. Helpfully, Exist tells me what correlates with these bad events. It would be even more helpful if Exist used causal inference to tell me what factors likely caused my bad events, telling me what to do to improve my mood or sleep better. In the last decade or so, developments in causal inference have allowed people to find causal connections just by looking at observational data, so the data Exist has should be sufficient to make such determinations.

16 votes

Tagged as New feature

Suggested 26 January 2020 by user Thomas

  • Sign in to comment and vote. Sign in by email
  • 26 January 2020 Thomas suggested this task

  • 28 January 2020 Josh Sharp approved this task

  • avatar

    I’ll have to look into how viable this would be, but conditionally approving the suggestion until I have a clear answer.

    28 January 2020
  • avatar

    Related to this: For cases where the data isn’t enough to infer the direction of causality, perhaps Exist could suggest experiments the user could perform to help identify causality.

    For example, suppose I work out some evenings, and I also track how awake I feel every morning. If there’s a positive correlation between nights I worked out, and feeling wakeful the next morning, then maybe the exercise caused the wakefulness. Or maybe some third factor caused me to feel motivated to work out, and caused me to feel wakeful the next morning. A simple experiment could help me distinguish between those two.

    26 April 2020