Hello,
I'm trying to estimate a 3-equation model in which 2 endogenous variables are continuous and the third (main interest) is ordinal.
I'm following Hotchkiss (2003) and using Nelson and Olson (1978) method of 2 Stage Limited Dependent Variable, in which the first stage is estimated for the two continuous variables and then their predicted values are used to estimate in the second stage.
The problem off course, is that when estimating and using the prediction, the standard errors of the coefficients of the second stage are clearly biased, since they rely on the predicted values rather then the observed.
Is there any procedure, like the TSLS command which correct the standard errors, that I can use?
Thank you,
Idit.
I'm trying to estimate a 3-equation model in which 2 endogenous variables are continuous and the third (main interest) is ordinal.
I'm following Hotchkiss (2003) and using Nelson and Olson (1978) method of 2 Stage Limited Dependent Variable, in which the first stage is estimated for the two continuous variables and then their predicted values are used to estimate in the second stage.
The problem off course, is that when estimating and using the prediction, the standard errors of the coefficients of the second stage are clearly biased, since they rely on the predicted values rather then the observed.
Is there any procedure, like the TSLS command which correct the standard errors, that I can use?
Thank you,
Idit.
Comment