So if each section has the same grade distribution, if you finish in say the top 1/2 of your section, you finish in the top 1/2 of your entire class most likely. If you receive a B, and 50% in your section get Bs or better, the same should apply to all sections, thus meaning you're still in the top 50%.

Other than the fact that the competition might be tougher due to people either working harder or just being smarter, thus having been offered scholarships, it wouldn't really matter if your rank is calculated against just your section or the whole class.

Or am I missing something here?

An example.

School X has a class size of 200 students, and it awards 100 students scholarships (50% of the class) that require each student be in the top 50% of the class for renewal. If it has four sections, it can stack like this:

Section 1 (Stacked - all students received scholarships)

50 students = 25 students are above the median

*Section 2* (Stacked - all students received scholarships)

50 students = 25 students are above the median

Section 3 (Unstacked - no students received scholarships)

50 students = 25 students are above the median

*Section 4* (Unstacked - no students received scholarships)

50 students = 25 students are above the median

Each class is graded on a curve, so let's say 25 students from each class are at or above the median. All sections use the same curve. At the end of the year, the GPAs for all students are combined and the class rankings are determined. The top 50% of each section will comprise the top 50% of the class. So by stacking, the school ensures that no more than 50 of the original 100 students awarded scholarships will be able to renew. Without stacking, it would be theoretically possible for all 100 students to renew.

I'm not saying schools actually do this, but this is the theory of stacking.