Why are so many EARNWEEK values zero?

I am trying to create earnings equations with March 2009 data but alomst 35% of the cases in outgoing groups list zero earnings, even after eliminating self-employed people. Is there something obvious I am missing here? Thanks.

You are limiting the population correctly, however, because March samples include an oversample (pulled from months other than March) some individuals with a code of MISH==4 or MISH==8 may be from other months and therefore not included in the earner study for March. If you applyEARNWT, these people will automatically be removed from the analysis and your frequencies should look more appropriate. If you would prefer to look at unweighted frequencies as well, you can drop cases with EARNWT==0, and only persons who were a part of the Earner Study will be left.

I hope this helps.