Testing for heterogeneity in panel data

[Jordan Smith]

I am attempting to test for individual specific heterogeneity in panel data and thereby show that OLS regression is inconsistent and bias and therefore use either a random or fixed effect model.

I am having some issues finding out what the command is for such a test is. I believe a sdtest may work however I am not sure how to specify this command in Stata.

Thank you in advance for your assistance.

[Clyde Schechter]

One way is to run -xtreg-, and then follow that with -xttest0-. The latter will test the heterogeneity of the random effects.

Then you can try -xtreg, fe- on the same model. The last line of the output from that command is an F test of the heterogeneity of effects at the panel-unit level.

(And if you want to test adequate consistency of the -re- vs -fe- models, there is -hausman-.)

See online help for all of these commands.

[Nazzarena]

How would do you do the same in hierarchical models with longitudinal data? E.g. if you have repeated measures for individuals nested in villages, how do formally test for heterogeneity by village?
Thanks