Hi all, I'm conducting a diff-in-diff analysis and I'm using margins to test the parallel trends assumption. My outcome variable is "medcost" (0,1) and I've included an interaction term for year (2009-2012, 2016-2019) and treatment (0,1). Here is my code: logit medcost i.year##i.treat2 if _age65yr==1 & mcaid_elig==1 & coverage==0 [pweight=_llcpwt], cluster(_state) allbaselevels

The output omits 2009 and 2010 and uses 2011 as the base level.

Thank you in advance for your help and advice!