Probit regression with minimum and maximum values of primary predictor variable

[Magnus Carstens]

Hi guys,

I would like to know the difference in probability of my main primary variable causing an event on my dependent variable as it is held at it's minimum and maximum values.

Does anybody know how the command for this in Stata? I'm using version 13.

My probit regression looks like this:

meprobit lead_fail_ucex LJI_lag i.defacto_lag i.legselec_bin_lag mil_gdp_l_lag civil_society_lag gdp_log_lag i.regimetype economic_control_lag gdp_growth_lag naturressourcer_log_lag gdp_log_lag diffusion_lag cold_war duration_leader duration_leader_sq duration_leader_cub time time_sq time_cub || id_ctry: , vce(robust) intpoints(12)

The minimum and maximum values for my main predictor variable LJI_lag are 0.012 and 0.955 respectively.

I'm a stata novice so all your help will be greatly appreciated!

Best regards,
Magnus

[Clyde Schechter]

margins, at(LJI_lag = (0.012 0.955)) pwcompare

[Magnus Carstens]

Thank you very much!

Do you also know how I can hold all the other covariates at their means? And margin corresponds to coefficient in the regression, right?

[Clyde Schechter]

margins, at(LJI_lag = (0.012 0.955)) atmeans pwcompare

The default output of -margins- after -probit- is the predicted probability for the configuration of predictors specified. The -pwcompare- option takes it one steps farther and calculates the differences among all of those, which gets what you asked for in #1.

I don't know what you mean by "margin corresponds to coefficient in the regression."