Let's think about this for a second. We have to assume that, as a test taker, he was already trying as hard as he could to "guess" (i.e. select without absolute certainty) the correct answers to all questions all the time, and that even so he scored a 33 on the first test. Provided he was still "guessing" as hard as he could on the second, we can attribute any increase in percentage of "guessed" questions answered correctly to chance. The scale for AAMC 11 provides that there would be a difference of at least 14 questions between scoring a balanced 33 (11 PS, 11 VR, 11 BS) and an even 42 (14 PS, 14 VR, 14 BS). Being generous, let's assume that for every question he "guessed" on, he had narrowed down the choices to two answers. This gives a probability of (0.5)^14 = 0.000061, that is a 0.006% chance that he "guessed" his way from a 33 to a 42.
Aside from this, I'm highly doubtful any legitimate adcom would actually interpret the score increase like this and tell an applicant something so insulting.