Drop observations if not meeting frequency criterium

[NORMAN DAVELAAR]

My dataset consists of time-series observation of US firms with identifiers like gvkey, cusip and a construct variable cusip_fiscalyear. I use data between 1995 and 2005, but need to drop all observations of a particular firm if the observations are less than 8 years, that is if for instance gvkey appears 8 times or less. I suppose I need to generate a kind of frequency variable? And I suppose I can use any one of the three identifiers.

[Øyvind Snilsberg]

if the data are without gaps,

bysort gvkey: drop if _N<8

[Nick Cox]

FWIW, the plural criteria is of Greek origin and its singular is criterion.


A criterium is a kind of cycle race.https://www.merriam-webster.com/dict...rmal%20traffic I didn't know this until I just looked it up.

[NORMAN DAVELAAR]

I already found the solution using the bysort command

[NORMAN DAVELAAR]

Thx Mr. Snilsberg

[NORMAN DAVELAAR]

....and Mr. Cox for correction of word usage.