Reporting Open Restaurants for King County

[EmilHafeez]

We generate the entire SC sample including King County such that there are control units available whenever the treated units have data available. The regressions are conducted on the relative months where there is common support between the two groups (so, restrict the generated control restaurants to when there are treated units). So, there are always control units when we want them; the 500 donated units for the SC units have complete data until 24 months after implementation, so postX = 1.

Relative-month respects differentiated entry points for King County (229 and 236 for IS and DT). For one of its occasions, data only runs till +6. For this occasion, the treated unit has a post8 of 0. For the other occasion, the treated run till ~+13 so post8 should be 1 for those.

However, for the analysis, we restrict to the treated's observed months. So, in the case of King where the treated post8 is 0, do we want it to show 1 or 0? I can argue both ways.

[EmilHafeez]

[EmilHafeez]

presented and closed -- solution decided in teh analyst meeting and in the ML weekly meeting was to replace the post8 variable with post6 and then show the integer values for King county accordingly. I've made this change in the men0labeling-prop-score-and-sc-combined script and @lloydheng has also made it in the script that generates synth_full,synth_drive,etc. Solution tested and is working, appendix table 1 updated