Step 1: Looking Backwards through IV, TWFE and OLS vs Looking Forward Towards the Target Parameter

All models are wrong but some are useful, so said George Box. But wrong in what sense? And useful for what? This substack will share some thoughts about an ongoing topic I’ve been having for a long time now — step 1 in my diff in diff checklist called “Define your target parameter”.

In today’s substack, I’m going to back up though and introduce the notio…