Hi all - I am using CEM in stata and have a quick question. I am using four variables in my dataset for matching and using CEM weights in the regression model as control. When I include the four matching variables in the regression as well, I find that the two of the variables are significant. Is that odd that these variables are significant even though they were used for matching/balancing? Let me know what could be going on or if you need more details. Thanks, Rajat.