Dear all,
I need help regarding interpretation of results in bivariate random effects probit model estimated by user-written cmp command by Roodman (Roodman, D. 2011. Estimating fully observed recursive mixed-process models with cmp. Stata Journal 11(2): 159-206.) Please find attached the results after using union dataset.
lnsig_1_1 is the ln of st. deviation of random effects for first equation, and lnsig_1_2 for the second equation
atanhrho_12 is correlation between erros of equation, arc-hyporebolic tangents.
I am not sure the difference between atanhrho_12 and atanhrho1_12. When I do not control for idcode (i.e no panel structure is taken into account) there is only atanhrho_12, which should be arc-tang of correlation between errors of equations.
Could you give me more information about output Random effects parameters and Level: idcode and Level: observations?
There are four estimates that I need to report together with significance level: rho (correlation between errors), rho_alpha correlation between random effects and sigma alpha1 sigma alpha2 (st. deviation of random effects).
I need to test the hypothesis of independence in the errors and the random effects of the two equations, and I am not sure which values I need exactly to report and comment. Also, in the part of the output Random effects parameters, I have no test nor significance level.
Thank you,
Best regards,
Aleksandra
I need help regarding interpretation of results in bivariate random effects probit model estimated by user-written cmp command by Roodman (Roodman, D. 2011. Estimating fully observed recursive mixed-process models with cmp. Stata Journal 11(2): 159-206.) Please find attached the results after using union dataset.
Code:
webuse union, clear cmp (union = age south black || idcode:) (not_smsa = age south black || idcode:), ind($cmp_probit $cmp_probit) tech(dfp) redraws(5, anti) nonrtolerance
atanhrho_12 is correlation between erros of equation, arc-hyporebolic tangents.
I am not sure the difference between atanhrho_12 and atanhrho1_12. When I do not control for idcode (i.e no panel structure is taken into account) there is only atanhrho_12, which should be arc-tang of correlation between errors of equations.
Could you give me more information about output Random effects parameters and Level: idcode and Level: observations?
There are four estimates that I need to report together with significance level: rho (correlation between errors), rho_alpha correlation between random effects and sigma alpha1 sigma alpha2 (st. deviation of random effects).
I need to test the hypothesis of independence in the errors and the random effects of the two equations, and I am not sure which values I need exactly to report and comment. Also, in the part of the output Random effects parameters, I have no test nor significance level.
Thank you,
Best regards,
Aleksandra
