The number of subjects reported in the Dimensions table can have little to do with the physical number of subjects in your dataset. The number gives you an idea as to how efficient PROC MIXED can be in processing your data. If PROC MIXED can identify a common subject across your RANDOM statement(s) and REPEATED statement, then PROC MIXED will report the number of subjects as represented by that common subject effect. If not, then PROC MIXED will report a 1. If PROC MIXED cannot identify a common subject, then the procedure cannot block the data when estimating the model and the modeling process is less efficient.

PROC MIXED reports the number of subjects as determined by both the RANDOM and REPEATED statement. Both of these statements have SUBJECT= options. If the SUBJECT= option is not used, then the number of subjects coming from that statement is 1. PROC MIXED reports the number of subjects as the minimum of the subjects from each of the RANDOM and REPEATED statements. ****If there are multiple RANDOM and/or a REPEATED statement and the subject effects are not nested then PROC MIXED reports only 1 subject in the dimensions table. ****

The number of subjects is not used in determining any statistics in the subsequent analysis. Specifically, calculation of F-statistics and degrees-of-freedom calculations do not depend on this subject value.