psmatch2 provides graph of Covariate Balancing Test without covariates

[karthick veerapandian]

Hello Forum members,

I am facing a problem in PSM method. Although I am aware of the recent development in kmatch, I would like to work on psmatch2 command just for the purpose. When I run the following codes

psmatch2 migrant rural sc st bc mbc dnc_new male mar_current mar_oth primary middle high hisec diploma graduates oth_edu hh_memb self_agri self_nonag cl_agri cl_ind cl_serv salary_pvt other_work hh_rent Pucca Semipucca marginal small semimedium medium large highinc upmidinc lowmidinc, outcome(total_income_yearly) logit ate n(1)

pstest rural sc st bc mbc dnc_new male mar_current mar_oth primary middle high hisec diploma graduates oth_edu hh_memb self_agri self_nonag cl_agri cl_ind cl_serv salary_pvt other_work hh_rent Pucca Semipucca marginal small semimedium medium large highinc upmidinc lowmidinc, graph both

I am getting a graph of Covariate Balancing Test without covariates name as follows.
Graph.gph

I am surprised to see this as for my another analysis I used same command with different covariates but used to get this graph with covariates. Can anyone guide me where I went wrong and how it could be resolved it?

[karthick veerapandian]

can anyone help me to resolve my issue?

[Felix Bittmann]

Just some wild guesses: Maybe the number of covariates makes a difference? What happens if you remove some, does it work then? If so, I would gradually increase the number of covariates and see when it breaks down. I also had the issue that some programs collapse if varnames or varlabels are too long. Since you have a case where it works, and one where it does not, I would attempt to gradually change the model to find the bug.