I haven’t had luck finding Pearl’s 6-step approach for determining minimal set of variables (illustrated by Shrier & Platt. Reducing bias through DAGs. BMC research Methodology 2008 8:70 that I found in a bibliography. Can you help a brother out?
Thanks! The title is unexpected. How did you find this, if I may ask? Just so I can learn from the process!
I like to experiment with different ways to formalise my thinking. Recently I’ve been been learning about DAG’s, conditioning, exchangeable exposures and such to design experiments. It’s certainly helping me make my ideas clearer.
I was reading a presentation made by a university person online that cited that article and thought the idea of ‘determining a minimal set of variables’ would be an intractable problem. I don’t see how someone could, in the abstract, determine a function or something that would tell someone the minimal set of variables to put in a DAG.
Having no scanned through the article, there is so much that is beyond what I’ve seen that I don’t expect I’ll be able to make sense of this article for a while. Moreover, I seem to have misinterpreted the content of the paper based on the simple sounding title I saw in the presentation!
I don’t see how someone could, in the abstract, determine a function or something that would tell someone the minimal set of variables to put in a DAG.
That is not the problem that this paper tries to solve. The paper assumes you know the graph, and are trying to find a sufficient set of variables to condition on to get d-separation.
To determine the minimal set of variables to include in the graph, you generally need subject matter expertise, ie external causal knowledge. Essentially, you need to be able to claim that there does not exist a variable not in the graph which is a common cause of two variables that are in the graph. (With a faithfulness assumption you may also be able to remove certain variables based on the data)
hello LW elders!
I haven’t had luck finding Pearl’s 6-step approach for determining minimal set of variables (illustrated by Shrier & Platt. Reducing bias through DAGs. BMC research Methodology 2008 8:70 that I found in a bibliography. Can you help a brother out?
http://ftp.cs.ucla.edu/pub/stat_ser/r254.pdf
Why do you care, btw?
Thanks! The title is unexpected. How did you find this, if I may ask? Just so I can learn from the process!
I like to experiment with different ways to formalise my thinking. Recently I’ve been been learning about DAG’s, conditioning, exchangeable exposures and such to design experiments. It’s certainly helping me make my ideas clearer.
I was reading a presentation made by a university person online that cited that article and thought the idea of ‘determining a minimal set of variables’ would be an intractable problem. I don’t see how someone could, in the abstract, determine a function or something that would tell someone the minimal set of variables to put in a DAG.
Having no scanned through the article, there is so much that is beyond what I’ve seen that I don’t expect I’ll be able to make sense of this article for a while. Moreover, I seem to have misinterpreted the content of the paper based on the simple sounding title I saw in the presentation!
That is not the problem that this paper tries to solve. The paper assumes you know the graph, and are trying to find a sufficient set of variables to condition on to get d-separation.
To determine the minimal set of variables to include in the graph, you generally need subject matter expertise, ie external causal knowledge. Essentially, you need to be able to claim that there does not exist a variable not in the graph which is a common cause of two variables that are in the graph. (With a faithfulness assumption you may also be able to remove certain variables based on the data)