[Clyde Schechter]

If you think those are important, you can get them by abandoning fixed effects regression. Look into using a random-effects model, or, perhaps best of both worlds, -xthybrid-. The latter is available from SSC.