You can query a lot of information about SAS datasets, variables, and other objects using the DICTIONARY "tables". In particular for this one the SQL code is querying the DICTIONARY.COLUMN metadata that lists all of the variables.   To test it first run that code that creates the GOOD and BAD datasets. Then run the PROC SQL step that queries the metadata and check out the results. Or just run the PROC SQL code and change the WORK to the name of the libref you want to check instead. Remember that values in LIBNAME and MEMNAME fields in DICTIONARY.COLUMN are always in uppercase only. ... View more

@Sudeep_Neupane wrote: Makes sense. It seems like I will have to SET then as it matches your first scenario. Well basically, I have four waves, and within those waves, the same variable is calculated 5 to 6 times at different periods of time. Depending on how it is stored you might have a series of variables.  X1 - X6 for example. Or you might have multiple observations per wave with the time period as a secondary key variables.  But that is probably not how it is currently setup based on the observation counts you provided before.   You can always convert from one structure to another structure depending on your reporting/analysis needs. ... View more

It works for all text files that can be read with data steps. ... View more