GWTC3-event-review.md

@@ -2209,7 +2209,20 @@ grav peak not 0
 </table>
 
 ## A posterior checks
+Visually inspect the A posteriors to see if everything looks OK. Things to look out for: is GR recovered (within or close to 90% CI represented by vertical lines); weird posterior shape (like double peaks); how it compares to gwtc3 posterior (about the same, better, worse).
 
+Plots last updated 7th November 2023.
+
+Potentially problematic events (excluded from gwtc3 analysis): GW191109_010717, GW200115_042309, GW200219_094415, GW200225_060421, GW200316_215756, GW190521_030229, GW151012_095443, GW170729_185629
+
+I propose following key for filling in the table (you can include brief text note after the symbol if you want to elaborate). Use 2 symbols. 1st represents how bilby posterior looks:
+* :white_check_mark: nice single peak and GR recovered [`:white_check_mark:`]
+* :ok: GR recovered but posterior looks weird (2nd peak, very broad, etc.) [:ok:]
+* :x: GR not recovered (A=0 significantly beyond 90% CI) [:x:]
+2nd symbol represents how well does bilby posterior compare to gwtc3 one:
+* :arrow_up: bilby has noticebly better posterior (narrower, 2nd peaks damped/removed, closer to GR) [:arrow_up:]
+* :arrow_down_small: bilby is noticeably worse (wider, more 2nd peaks, farther from GR) [:arrow_down_small:]
+* :neutral_face: both posteriors look more or less the same [:neutral_face:]
 <table>
 
 <tr>