Retention in engineering and physics is really bad. Only about 30% – 40% of students who start in engineering or physics get to their degree in that field in six years. Students often go other routes after one of their first STEM courses. There are a lot of reasons for this. And this is something I care a lot about, but not the topic of this blog post.

This blog post has to do with curving. I wonder if our upper level students are getting screwed over.

Here’s the thought experiment. Consider teaching the first class of a STEM series like physics I or Chemistry I. Your students do okay, and you as the instructor use a curve for your class. You implement this reasonable curve

Students in the red section receive an F, Orange D, Yellow C, Green B, and Blue A. This curve has a healthy B section!

But what happens when we think about retention. We know that only 30% of students are going to make it. So if we have a class of 100 students. That means only 30 will make it to the upper level classes.

We are going to make a bunch of assumptions:

* The students that don’t make it come from the lower end of the class. (Not always true)

* Students are always working as hard as they can on any given class. (This is a wildly untrue assumption)

* Students don’t vary in their placements in class (So this model is the Temperature = 0 case. Unfortunately this is usually close to true too.)

If we lose 70% of the population of the orginal first class and we decide to curve in a course like Modern Physics or Oraganic Chemistry. Are we selecting from the top-tier elite? Let’s replot that curve but now drop the 70% of students.

The black section refers to the 70 students that didn’t make it up upper level status. The color ranges for the curve of the 30 or so students we have left. Notice that even students that were in the A range in that initial class are now in the C/D range in the upper-level class.

Should students GPA be rewarded for being successful in these harder classes, similar to how an honors or AP is often weighted differently than regular classes? I don’t think so, but it is interesting to think about?

Of course determining the grade a student gets is a complicated process and is set by many factors. Typically broad curves like this aren’t applied but it always in the back of my head when applying a curve for a given class.