Hot take: today, little consensus on how to do #causalinference @yudapearl emphasizes “qualitative” graphical model based approach @blei_lab emphasizes latent factor analysis for ‘surrogate causes’ @StatModeling: linear regr for “quantitative” causal discovery 👉other results ! pic.twitter.com/oqzdFy30t4

I am pretty sure that @yudapearl and @StatModeling mean the same thing and just use different words and approaches (i.e. you can use @yudapearl approach to show that @statmodeling does it right).

I am not so sure. Why? Because all my attempts to convince @statmodeling to show us how he solves a simple causal problem (one whose solution is known in advance) ended up in failure. I lost my charm. Simpson's would be a good example #Bookofwhy

Have the two of you discussed it in person? I am, quite honestly, surprised. I find both of you to generally be very precise. Maybe it is the issue that you two use different language? Could you give an example of a paper of @statmodeling where you would disagree?

In person? You must be kidding. If you knew the number of hours I spent on his blog, you would not say "different language". The example that brings everything to the surface is Simpson's paradox. I dont have the strength to dig it again from his blog. If you can, please share.

this? statmodeling.stat.columbia.edu/2014/04/08/und…

But upon reading it, I am not sure you two are really disagreeing. The strongest he says is "there are also non-causal questions I find interesting" which seems perfectly fine no?