3 ms·
Number of data points aside, I would like to see one causal model worked out in depth. Do you have a sample link?
by hsjqllzlfkf 3y ago
Number of data points aside, I would like to see one causal model worked out in depth. Do you have a sample link?
- nerdponx 3y agoHow about a book and lecture series full of them? https://xcelab.net/rm/statistical-rethinking/ https://xcelab.net/rm/statistical-rethinking/
- hsjqllzlfkf 3y agoThank you! I've read one of Judea pearl's latest books (called I think "the question of why") and while I found it extraordinarily convincing to make the case that the "data first" approach to inference is fundamentally flawed, I also couldn't extract anything quantitative from his graphs. Maybe my fault, it's on my to-reread list. Thanks for the link.
- nerdponx 3y agoPearl is the author of a more technical book called Causality, which is where all the quantitative stuff is. But McElreath's material on causality is a distillation of some of Pearl's (and others') work, so start there.
- hsjqllzlfkf 3y agoI'm halfway thru the second lecture, and what an absolute find :-)