Hi all. I have a large dataset that is sorted by year, then by timestep, then by score (descending). Each timestep has 16000 obs, but I only want the highest 4000 scores for analysis. Is there a way to either: 1) count the obs by year by timestep, adding a new var called count, and then I can later drop all obs whose count is greater than 4000; or 2) write a loop that counts and moves to a new dataset the top 4000 obs of each timestep of each year?

Thanks a lot. Phil