Performance on one test form, during one administration, can be compared validly with performance on another test form taken at a different administration. We accomplish this by using classical equating theory from the field of psychometrics.
Test forms may be used in subsequent administrations. That's one reason for the strict security measures we employ. In each administration, a random sample of examinees, of sufficient size, will be administered test forms previously administered in an August administration, and another random group will be tested with test forms administered in an April administration.
Most repeaters are those who did not perform as well as they thought they should. However, our studies of repeater performance show that there is only a slight score gain for repeaters on the whole. This indicates that the test is reliable.
A much larger number of August examinees are repeaters compared to April examinees. As a result, the mean raw scores for August test-takers are lower than for April test-takers. Therefore, in scaling the scores, this is taken into account."
~John Hackett, Associate Director for MCAT Operations, AAMC, responded to my enquiry concerning this as follows.