Hunt is right in #57, although both the effect that I mentioned and the effect that he mentioned operate at the same time, on top of each other.
For example, in a three-quarter class, if the people who scored 65% overall in the first quarter, moved on to the second quarter, and changed nothing, their predicted score for the second quarter would be 42.45%. They might cut their losses at that point, rather than continuing to the third quarter where their projected score is 27.5%. If they were still in the class, it would make the student who started at 80% and has now dropped to 51.2% look better.
Persevering without changing anything is unlikely to work well, if things weren’t going well at first. Recognizing the need for change and making it is not super-easy.