A great paper on the intersection and/or tension between traditional econometrics and modern causality.

economics.mit.edu/sites/defaul...

One of the most underrated concepts in observational causal inference is sensitivity analysis. Sensitivity analysis helps you shift the conversation from “no confounder exists” to “it's unrealistic a confounder of strength X exists”. That’s a much more defensible claim.

#CausalInference

With Causal Inference, we often draw Causal DAG's. If this DAG is incorrect in relevant parts, our analysis will be off. Unfortunately, we can never fully validate our Causal Graph. How do we deal with this uncertainty? Read our latest blogpost:

medium.com/@causai_busi...