Hi StataList,
I have been trying to run diagnostic tests for an IV regression with an endogenous binary variable. In particular, my model is specified as
Y=a+bX1+cD where D is the endogenous dummy, and
D=k+fX2+gX3.
I am using survey data with pweights in my svyset.
Now, my problem is as follows:
I intend to first run a regression of D on my exogenous variables X1 and the instruments X2 and X3. Then, I predict the residuals from the equation and regress Y on the residuals to test for endogeneity. I will then continue to test for instrument relevance and instrument exogeneity. Here is my code:
svy, sub(sample): logit drink i.white i.male age c.age#c.age i.ftptwk ib13.edlev00 i.dvmrdf2 i.depchild i.govreggb i.immigrant i.cigsmk i.freqmtf c.faminc
predict res2, residuals
However, -predict- followed by -residuals- gives me an error message; it seems I cannot obtain a residual estimation using both logit and svy.
Please help, I really need to solve this endogeneity issue!
I have been trying to run diagnostic tests for an IV regression with an endogenous binary variable. In particular, my model is specified as
Y=a+bX1+cD where D is the endogenous dummy, and
D=k+fX2+gX3.
I am using survey data with pweights in my svyset.
Now, my problem is as follows:
I intend to first run a regression of D on my exogenous variables X1 and the instruments X2 and X3. Then, I predict the residuals from the equation and regress Y on the residuals to test for endogeneity. I will then continue to test for instrument relevance and instrument exogeneity. Here is my code:
svy, sub(sample): logit drink i.white i.male age c.age#c.age i.ftptwk ib13.edlev00 i.dvmrdf2 i.depchild i.govreggb i.immigrant i.cigsmk i.freqmtf c.faminc
predict res2, residuals
However, -predict- followed by -residuals- gives me an error message; it seems I cannot obtain a residual estimation using both logit and svy.
Please help, I really need to solve this endogeneity issue!
