I looked this all up once (if you could search the forum a year or so back, you might find my previous posts). There is a version where only 25 or so questions are administered. I think there might be situations where up to 35 are, but some of those don't count into your score, if I recall correctly. The test resets to your grade placement if you haven't taken it in a certain number of months--so often, the fall administration has reset to the default starting points, which is items that are slightly below your nominal placement. On the next administration, the starting point is based on your previous score. If you have a third administration that year, and you happen to have scored really well on the second administration, you might run into artificial ceilings, because you already used up most of the high-level items in the item pool, which now can't be repeated for some interval (a year?).
Another factor is the intelligent item selection algorithm, which can be really thrown off by an early error, such as sometimes occurs when young children haven't settled into the testing process yet, or otherwise don't find it sufficiently engaging for full attention. Since the probe is so short, you may not have enough items to work your way back up to the challenging items before the test ends.
And here's the guidance from NWEA on who should take the 2-5 versus the primary version of MAP:
https://www.nwea.org/content/uploads/2015/01/MPG-to-MAP-Transition-Guidance-Document-JAN15.pdfNote especially the absurd skyrocketing of SEM after about RIT 190 in reading or 200 in math.
