I would follow up on Patrick's comment and ask whether you are use the longitudinal component of your data in some sort of hazard rate model or something where you are just using the length of time in a state in something like a categorical analysis of some sort. Either way, flipping the time horizon from rows to columns is NOT the format the SAS PROCs that I can think of would use as an input dataset. If you are still insistent on having all of these variables, there have been several decent suggestions for doing the transformation and promote one of the responses as the right answer. (You should also read the finer options and examples on PROC TRANSPOSE. It probably IS what you want to use.)
... View more
