xthreg and missings

[Ernesto Cassetta]

Hello stata users,
I am trying to use the command xthreg for a balanced panel, where the dependent variable and most of the independent variables (except a few) have no missings. Having the confirmation that the panel is "strongly balanced" after running "xtset id year", when I run xthreg, Stata gives me the error "Panel threshold model need balanced panel, check your data!". And this just because of a few independent variables which have some missings. When I exclude them, the xthreg command works. Now, reading previous posts here, some users suggested to drop directly the missing observations; however in my case it does not make sense because dropping observations because of just some independent variables would lead me to loose information. Any idea on how to proceed with this?
Thank you.

[Andrew Musau]

however in my case it does not make sense because dropping observations because of just some independent variables would lead me to loose information.

Even though most estimation commands do not explicitly state that they exclude observations, the vast majority implement listwise deletion. This is equivalent to dropping all observations with missing values in at least one of the variables. If the missing observations are random, this is not a problem. However, it becomes an issue if the missing observations are non-random.

[George Ford]

If you want to keep all the data, then you'd use Stata's impute commands.

[Ernesto Cassetta]

Thank you guys.