Calculating Odds ration in probit model

[Kazi Aiman Udoy]

Hello Everyone,
Hope you are all doing well

I am doing a probit model. My commands are working properly without any error.

My command is:

probit vocationaltraining hhsize gender age agesq literacy uneduc_adults no_educ primary_edu pscpass jscpass sscpass hscpass tertiarypass ln_savings ln_asset ssn ffp urbrural mari_status barisal chittagong khulna rajshahi rangpur sylhet members_in_agri agrishock, robust

margins, dydx(*) post

Now, I want to find the odds ratio for my probit model. What is the command to to that?


Appreciation in advance for the help.

[Joseph Coveney]

Now, I want to find the odds ratio for my probit model. What is the command to to that?

If you want odds ratios, why don't you just fit a logistic regression model?

help logit
help logistic

[Kazi Aiman Udoy]

Isn't it possible to calculate odds ratios in probit model? What if I run probit for coefficients and average marginal effects and logit for odds ratio? Will it be appropriate to use different model? Or should I use only one? In that case, are you prescribing to use logit for the three purpose?

[Rich Goldstein]

there is a very rough approximation between the coefficients of the probit and logit (note that this does not refer directly to the ORs): multiply the probit coefficient by 1.6 to get, very approximately, the log odds which you can then exponentiate to get a very approximate OR - this can be traced back to the (pi/sqrt(3)) difference in the variances of the models; see, e.g., Amemiya, T (1981), "Qualitative response models: a survey", Journal of economic literature, 19: 1483-1536; you are much better off directly using logit (with the or option) or -logistic-

[Kazi Aiman Udoy]

Thank You. I decided to use logit.