Have you ever looked over a SAS® dataset or report you’ve prepared and concluded that it “looks good” or worse still, “good enough”? One of the great advantages SAS programming – the ease with which you can manipulate data – is also a potential hazard. It’s far too easy to create something that’s close, but not exactly correct. Beyond instinct or some sort of sniff test, how do you know that your results are correct? How can you prove it to others? As SAS programmers or users, we should be striving for clear and accurate data, not just “good enough” data.
A SAS log showing no syntax or data errors or warnings and some reasonable numbers of observations is only a small step. This paper offers a common-sense approach to simple data validation, expanding on some of the most frequently used validation techniques. Some may consider these recommendations obvious. Experience has proven that they are obvious only if you follow them.
conference:
Paper Type:
Paper
