The only way this could possibly work and be a “fair” comparison across students who take all sections at the same time and those who take only selected sections is to scale the scores of subsections differently. For example for an applicant who missed 3 questions on a subsection but took all subsections when scaled against all applicants who took the whole test may make a 35, but for an applicant who only took that subsection when compared to other applicants who only took that one subsection, 3 wrong could yield a 32. That would be fair. Kids in similar testing situations would be scaled against each other.