Hi,
I find that the "ivprobit" function tends to have difficulties when users want to explore interaction terms because the function does not allow interaction terms to appear. I would like to ask if it is possible to implement the exploration of interaction items in the control function?
My code is as follows, where x1 is the endogenous variable, x2 is the exogenous variable, z is the instrumental variable, and y is the dichotomous independent variable. c_x1 is the centered x1, c_x2 is the centered x2, and c_z is the centered z (I centered these three to make it easier to interpret the coefficients).
and the coefficient of c_x1#c_x2 shows the effect of interaction term.
Do you think it's plausible? Thank you very much.
Best regards,
Ke Ju
I find that the "ivprobit" function tends to have difficulties when users want to explore interaction terms because the function does not allow interaction terms to appear. I would like to ask if it is possible to implement the exploration of interaction items in the control function?
My code is as follows, where x1 is the endogenous variable, x2 is the exogenous variable, z is the instrumental variable, and y is the dichotomous independent variable. c_x1 is the centered x1, c_x2 is the centered x2, and c_z is the centered z (I centered these three to make it easier to interpret the coefficients).
Code:
reg x1 z x2 c_x1#c_x2 c_x2#c_z predict res, resid probit y x1 res x2 c_x1#c_x2
Do you think it's plausible? Thank you very much.
Best regards,
Ke Ju
