Dear Statalisters,

When pre-processing my data for a propensity score analysis I would like to make an analysis of the best matching method as is suggested by e.g. Gary King. Within one of his articles, he suggest that you use a frontier however as that option is not available in STATA I am trying to 'just' match my data in three different manners (MM, PSM and CEM) and then analyse the 'best' matching method by looking at the decrease in bias between matched and unmatched.

However, for MM I want to use -psmatch2- but I am a bit confused by its meaning as there is a command -psmatch2- and -psmatch2 depvariable, mahalanobis (covar)-. Is propensity score matching used under -psmatch2- and mahalanobis matching under -psmatch2 , mahalnobis()-?

Any tips on how to decide which matching method to us are welcome.

Thank you in advance and have a great day,
Linda