When did we allow "causal inference" to mostly mean "causal inference with observational data"? At least that's how it seems to me. As someone who has to think a lot about inference with experimental *and* observational data, I think this is short sighted.
Comments
but I'm salty
The same is not true in other fields.
See also solutions to the replication crisis in psychology
I've personally become an "inference to the best explanation" person and am not sure it's possible to infer causation without some combination of significant amounts of experiment, observation and theory.
Dunkler Klaus 😈💀⚒️
#causalinferenceishard
100% disagree on the second :) Not that I think it actually matters...but *the* distinction is manipulation of exposure. If you want to argue that that means is an observational study with manipulation, then words might as well not have meaning.
It's yet another perspective that makes it obvious that the hard line distinction between obs/exp studies is pedagogically harmful and brings more confusion than clarity.
“Observational is where you observe certain variables and try to determine if there is any correlation. Experimental is where you control certain variables and try to determine if there is any causality.”
https://academic.oup.com/book/3313
Epidemiologist's perspectives on causality are really a very tiny part of science.
Here's another one for folks to consider. Discussion of causal inference w/ agent-based models. A direct retort to some parochial views of causal inference one encounters in social sciences.
https://www.wiley.com/en-us/Agent-based+Models+and+Causal+Inference-p-9781119704461
In general, I think researchers (as a whole) still think that's not possible w/ observational data, so causal inference discussion is focused on that point.
this site is missing that level of curmudgeon.