Some SAS/STAT procedures display numerical results, graphical plots, or output data sets that contain observation numbers. For example, you might use the observation numbers in a regression diagnostic plot to identify extreme observations. Or you might use the observation numbers in an output data set for clustering results to identify observations in clusters.
As discussed in the section Observations Might Not Be Identified Reliably with Input CAS Tables, using observation numbers to identify observations is neither a good nor a reliable practice, especially when you use a CAS table as the input data set. Specifying ID variables in the ID statement (if supported) enables a procedure to use the ID variable values to identify or label the observations correctly.