Zero vs. Null

[GalacticAbacus]

Hello, I was doing a distribution analysis on a single variable. In this particular analysis, the null values as % of total was rather important, however due to calculation balancing, some of my null values went to zero. For the purpose of this study, all zero values are essentially null, however SAS does not see them as such and they get lumped in with my "values" when kicking out useful stats.

In looking around for a solution is seems most people are concerned with going from null to zero and not from zero to null. What is the most efficient way to transform my zero values to "missing"?

TS

[ballardw · Accepted solution]

In a datastep:

data want;

      set have;

     if var=0 then var=.;

run;

I would tend to create a new variable though in case I need the original values.