Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# st: Fwd: joint probability for all possible combinations using mvnp after mvprobit

 From Deepa Pradhan <[email protected]> To [email protected] Subject st: Fwd: joint probability for all possible combinations using mvnp after mvprobit Date Fri, 4 Oct 2013 12:06:40 +1000

```Hello

I am trying to get joint probabilities for all possible combinations
for a multivariate probit model.

I have three dependent variables (drcrops mixcrop and irrigators) and
I am using mvprobit and mvnp after that. But the results that I am
getting from mvppred and egen mvnp does not match (checking before
doing estimations for other combinations). If I could know what
mistake I am making, it would be great . How can I do it for other
combinations like Y1 only or Y1 and Y2 and so on.

Steps that I have followed are as follows for example for combination
where all dependent variables equals 1:

mvprobit (drcrops= hu  caste totland treat) (mixcrop = hu caste
totland treat) (irrigators= hu caste totland treat),draws(50)

mvppred pall,pall

mvppred xb, xb

mvppred stdp,stdp

matrix accum V= stdp1 stdp2 stdp3

matrix V=corr(V)

matrix C=cholesky(V)

generate k1=1

generate k2=1

generate k3=1

mdraws,dr(250) neq(3) prefix(z)random seed (123456789)

egen pr_joint=mvnp( xb1 xb2 xb3), dr(250) chol(c) prefix(z) signs(k1 k2 k3)

list pr_joint pall1s