Balancing property not satisfied in propensity score matching using pscore

[Caoilinn Okelly]

Hi all, I had a propensity score matching question, for my thesis I am doing propensity score matched smaple of B Corps and non B Corps. I wish to examine their EBITDAs given controls such as age of incorporation, industry, country and size of the firm. I am using pscore and Im having trouble with the "balancing property not being satisfied". I have ran the programme twice, using size and log of size (number of employees) and I am getting different results for which variable blocks are not balanced and Im wondering if anyone has any ideas as to what I can do about this.
*the first picture is with logsize ( this photo has more balancing property issues with other variables) but the only thing I changed between the two is changing log of size to size (where in the second photo, the problem is with the variable size) And there are less issues in the balance of the sample when using size instead of log of size. Any ideas as to what I can do from here would be greatly appreciated.

With logsize



with Size


reg EBITDA Bcorp Size Age industry.D country.D
pscore $treatment $xlist, pscore (psmatch) blockid(block) comsup