egen mean - missing value for some observations

[Rochelle Zhang]

Dear Statalisters,

this might be a rudimentary question:

I use the following code to generate mean for each industry and year combination,

bysort industryid year: egen mean_xvar=mean(xvar)


xvar is missing for a subset of my sample, I know the above statement still works,

is it correct to say that mean is generated without considering the missing observations?

Thanks
Rochelle

[Nick Cox]

I don't know what alternatives you are imagining, but Stata does ignore missing values in this computation, unless all values in a given group are missing. . (Presumably, "missing observations" means observations with missing values.) That is, there is no interpolation, imputation, or other replacement.