Hello, I am trying to estimate the effect of private tutoring on learning outcomes (math ability). I suspect private tutoring is endogenous hence am using -ivprobit- and -biprobit- to run an IV estimation. My dependent variable is binary. Here is my code:

Code:
ivprobit math_binary $child $hh $village1 (tuition=my_IV) i.state_code, first cluster(village_code) nolog
margins, dydx(tuition) predict(pr)

biprobit (math_binary $child $hh $village1 i.tuition) (tuition=my_IV), nolog
margins, dydx(tuition) predict(pcond1) force  //force option gives result else error in Stata12
My question is whether the marginal effect computed after each regression is comparable? Naturally, I am interested in how availing tuition (tuition=1) affects the probability of being better at math (math_binary=1). Is the option -pcond1 correct after -biprobit- or should I use something else?