Dear all,
I´m running a probit regression, in which the dependent variable has been created as follows:
gen Emp = 0
replace Emp = 1 if gender==1&ocupation==3
replace Emp = . if gender==.|ocupation==.
Some of the explanatory variables are also binary variables. My problem is that if I generate all the binary variables including the third line of commands (replace "variable"= . "whatever" ==.) when running the probit regression i get "error 2000" and as i do not have any string variable i guess there should be some problem with the missing values. The regression works if some of my explanatory variables are created without the third line of commands, but I doubt that it is right to include it just in some variables and not in others. Could you explain me how should i deal with the missing data?
Thanks!
I´m running a probit regression, in which the dependent variable has been created as follows:
gen Emp = 0
replace Emp = 1 if gender==1&ocupation==3
replace Emp = . if gender==.|ocupation==.
Some of the explanatory variables are also binary variables. My problem is that if I generate all the binary variables including the third line of commands (replace "variable"= . "whatever" ==.) when running the probit regression i get "error 2000" and as i do not have any string variable i guess there should be some problem with the missing values. The regression works if some of my explanatory variables are created without the third line of commands, but I doubt that it is right to include it just in some variables and not in others. Could you explain me how should i deal with the missing data?
Thanks!
Comment