All that verbiage and the best he can do with respect to establishing causality is an offhand remark that causality is “heavily implie(d)”. Controlling for conf

All that verbiage and the best he can do with respect to establishing causality is an offhand remark that causality is “heavily implie(d)”. Controlling for confounders won’t necessarily recover estimates of causal impact. At the least I’d like to see something like an instrumental variables analysis.