As an expansion of @sustagens good solution:   The format approach is often the most flexible. If you add variables then when someone asks a question like: What if the children are broken out at age < 13 and 13 to 17? Or other age groups for the adults? Then you have to go back to the data step, create another variable and rerun everything. And then do it again when they ask yet another age grouping.  I have had as many as 18 different age-based formats at one time as different projects have different reporting boundaries for example.   The groups created by formats in this fashion will work with most of the analysis (I say most because I haven't run every single analysis proc), reporting and graphing procedures. So to add a category you 1) create the new format and 2) apply in the procedure call(s). And for a few procedures (Tabulate, Report, Means, Summary) there is an option call MULTILABEL format that would do a summary by 1-<13, 13-17 and 1-17 at the same time. ... View more