Methods of discovering causal effects: Difference between revisions

From Machinelearning
No edit summary
No edit summary
 
(3 intermediate revisions by the same user not shown)
Line 5: Line 5:
! Method/strategy !! Notes !! Example
! Method/strategy !! Notes !! Example
|-
|-
| Try to deliberately control as many other variables as possible || <ref name="shalizi-identifying-causal-effects">Cosma Rohilla Shalizi. [http://www.stat.cmu.edu/~cshalizi/uADA/16/lectures/23.pdf "Identifying Causal Effects from Observations"]. April 7, 2016.</ref> ||
| Try to deliberately control as many other variables as possible || <ref name="shalizi-identifying-causal-effects">Cosma Rohilla Shalizi. [http://www.stat.cmu.edu/~cshalizi/uADA/16/lectures/23.pdf "Identifying Causal Effects from Observations"]. April 7, 2016.</ref> || ?
|-
|-
| Randomized controlled trial || <ref name="shalizi-identifying-causal-effects" /> Michael Nielsen also talks about this in his tutorial ||
| Randomized controlled trial || <ref name="shalizi-identifying-causal-effects" /> Michael Nielsen also talks about this in his tutorial || There are lots of examples of this. It would be good to explain a concrete one. For now, I'll just link to [https://en.wikipedia.org/wiki/Randomized_controlled_trial#In_social_science this section on Wikipedia].
|-
|-
| The [[Do calculus|''do'' calculus]] || This allows us to infer causal effects just by observation. I think this one breaks down into further strategies. Three of them are listed in p. 30 of <ref>Stephen L. Morgan; Christopher Winship. ''Counterfactual and Causal Inference: Methods and Principles for Social Research''. 2nd ed. Cambridge University Press. 2015.</ref> ||
| The [[Do calculus|''do'' calculus]] || This allows us to infer causal effects just by observation (i.e. without conducting an experiment). I think this one breaks down into further strategies. Three of them are listed in p. 30 of <ref>Stephen L. Morgan; Christopher Winship. ''Counterfactual and Causal Inference: Methods and Principles for Social Research''. 2nd ed. Cambridge University Press. 2015.</ref> ||
|}
|}
==See also==
* [[Creation of causal graphs]]


==References==
==References==


<references/>
<references/>
[[Category:Causal inference]]

Latest revision as of 00:25, 15 January 2019

Methods of discovering causal effects: basically, after observing or interacting with the environment, how can we discover what causes what? When are we justified to claim causal effects (not just correlation)?

Method/strategy Notes Example
Try to deliberately control as many other variables as possible [1] ?
Randomized controlled trial [1] Michael Nielsen also talks about this in his tutorial There are lots of examples of this. It would be good to explain a concrete one. For now, I'll just link to this section on Wikipedia.
The do calculus This allows us to infer causal effects just by observation (i.e. without conducting an experiment). I think this one breaks down into further strategies. Three of them are listed in p. 30 of [2]

See also

References

  1. 1.0 1.1 Cosma Rohilla Shalizi. "Identifying Causal Effects from Observations". April 7, 2016.
  2. Stephen L. Morgan; Christopher Winship. Counterfactual and Causal Inference: Methods and Principles for Social Research. 2nd ed. Cambridge University Press. 2015.