Expanding chapters

[malcolmbarrett]

"Other approaches" chapter:

• ML its own chapter
• At least one chapter for IV, RD, DID. Maybe more

Add:

• interaction
• time-varying
• triangulation, generalization

If possible, condense:

• continuous and categorical exposures

[malcolmbarrett]

A note on interaction since we don't have an issue yet: probably less sample size for when you marginalize over it but you want to be careful of the curse of dimensionality even when not estimating the interaction term for sake of stability

[malcolmbarrett]

idea for time-varying data: use Disney data on a rolling basis, e.g. across a day where the previous hour posted predicts next hour actual