The Hausman test result suggests using a fixed effect. However, as the model has several problems, I run the robust Hausman test. The robust Hausman test results refer to the use of random effects. Which model shall I use then? Another question is whether the "xttestpanel all" command (after estimating the FE model using "xtreg") is an official STATA command and if I can rely on its results.
-
Login or Register
- Log in with
Comment