Message from Judea Pearl
Dear blogger,
A few causality-related papers have recently been posted on my website:
- R-357: J. Pearl, “On Measurement Bias in Causal Inference,” http://ftp.cs.ucla.edu/pub/stat_ser/r357.pdf
- R-358: J. Pearl, “On the Consistency Rule in Causal Inference: An Axiom, Definition, Assumption, or a Theorem?,” http://ftp.cs.ucla.edu/pub/stat_ser/r358.pdf
- R-359: J. Pearl, “Physical and Metaphysical Counterfactuals,” http://ftp.cs.ucla.edu/pub/stat_ser/r359.pdf
- R-360: J. Pearl, “If Oswald Hadn’t Used Counterfactuals, My Robot Would Have,” http://ftp.cs.ucla.edu/pub/stat_ser/r360.pdf
- R-361: S. Greenland and J. Pearl, “Causal Diagrams,” http://ftp.cs.ucla.edu/pub/stat_ser/r361.pdf
As usual, comments, reservations and objections are most welcome, and can be posted on this forum.
And may that every brilliant thought be put to some good cause.
Judea Pearl
I find the idea of using causal diagrams very appealing. However, most discussions I’ve seen have not been very grounded in how causal diagrams might be used in analyses and have been too theoretical. The above Causal Diagrams paper is certainly the clearest explanation of the theory that I’ve seen thus far.
However, I was wondering if the author or any contributors might be able to point me in the direction of a paper that illustrates causal diagrams in action. That is to say, are there any papers that illustrate the use of causal diagrams in empirical work? I am from the field of economics and would prefer applications from there, but am open to any other good examples in the social sciences.
Comment by Dana — February 21, 2010 @ 3:08 pm
Dana,
This paper is not about Causal Diagrams, it is about a theory of causal reasoning, part of which can be articulated effectively in diagrammatic form, free of the details that obscure right from wrong in the mainstream social science literature.
The book by Winship and Morgan discusses several social science applications using diagrams. But, more importantly, take any social science study that you admire, and you will see that (1) the underlying assumption are easier to understand with diagrams, (2) a large portion of the results could have been obtained by just looking at the diagram, and (3) the next question you want the study to answer can either be answered from the diagram or proclaimed “unanswerable”.
I would suggest for example problems that seek to estimate the direct and indirect effects between variables. How do we know if they are estimatable? Especially if we do not assume linear equations.? Or , how do we know that a variable Z is an instrument relative to a pair X and Y? Or, how do we know what we are “testing” when we compute degree of fitness for a model? These are the sort of problems that causal diagrams answer. The diagram is not actually “working,” it simply guides you into doing the right thing.
Comment by judea — March 1, 2010 @ 9:56 pm
I have been examinating out some of your articles and it’s nice stuff. I will surely bookmark your site.
Comment by cabinets — June 12, 2014 @ 3:21 am