Additional topics in calculating propensity scores

[malcolmbarrett]

I'm not quite sure where these should go yet, but I'm going to link them to the ~~calculating scores~~ estimands chapter

• [ ] Show examples of ATO under different simulated conditions of overlap (callout box)
• [ ] Discuss how improving the model for the ATO weights improves the balancing in other modes, e.g. ato -> better fit -> better ecdf. Also show weighted tables
• [ ] Sample size: finite sample bias and overlaps vs att/ate/att, effective sample size -> how many people in region of overlap?, simulation for sample size (or a callout box on this topic)

[malcolmbarrett]

I think we need an effective sample size section