Big advantages of using the format approach for this: 1) you don't have to add any variables, so no additional passes through the entire data 2) you can assign the format as needed for specific analysis 3) one format can be assigned to multiple variables that use the same rule 4) multiple formats to regroup data as needed, just change the format for the proc at run time. 5) as long as values of only a single variable are considered it can be much easier to construct then if/then/else logic and checks to ensure that one start value is treated the same all the time (exception: multilabel but that's a bit more advanced and only useable for some procs) 6) used with special missing assignments .A - .Z you can get descriptions for what that missing means such as not asked, original value out of bounds (and differentiate between high and low), instrument failure, skip pattern and a plethora of conditions. And the missing values are still excluded from calculations such as means. ... View more