Saturday, June 14, 2008

Predictable Grades

Profgrrrl blogged about being able to predict where her summer (grad) students would fall on the grad distribution for a test about 2/3 of the way through a short semester. In the comments, Belle (who had a wonderful recent article about nonsense in some AP history? exams on the grading table) asked how that might be done.

I think Profgrrrrl's observation was about the sort of informal approach many of us do, where we have learned to expect some continuity in both "A" and "D" performance throughout the semester, but I've gone a bit further as a teaching tool.

In one gen-ed class, I make a point of recording the quasi-midterm percentage grade I report to them after one of the exams. I put it in a place in my gradebook where I can easily compare it to the final grade, pretty much at a glance. It is useful to know that most students who are borderline (particularly on the pass line) improve their grade on the final exam - mostly because the first exam is harder than they expect, but they can get those kinds of questions right on the final if they change how they study for the class. I can use that to encourage them to keep improving.

In my first semester physics class, I have been doing a retrospective study for a number of years, building a histogram of sorts where I put the final grade in the class next to the numerical score on the first hour exam. A pattern emerges where most bad scores fail and most good scores do well, but there are always a few cases where a 100 gets an F or a 40 gets a B. (The latter is possible because I follow one of many standard grading schemes where the worst exam is either dropped or replaced by the score on the final exam. The former is possible because some students don't keep doing what got them the good grade on the first test.) The extremes tend to track pretty well.

The interesting result is that the most critical group of students are the ones in the C-D range. For them, what matters the most is what they do about their performance on the first test. Do they sit down and work it all out correctly the very next night? Do they figure out the relationship between exam questions and key topics covered in the homework and adjust their study methods accordingly? Do they realize that copying an example in the book or getting someone else to tell them what formula to use to get the homework correct is not quite the same as "doing" the homework with the goal of "learning" how to do problems on a test? If they do, their grades generally go up into the B-C range, sometimes even into the A range. If they don't, their grades generally go down into the D-F range. A person with a 70 on that first exam is almost literally sitting on the Continental Divide of performance, sliding down the razor blade of life.

Again, conveying this sort of information can make the act of returning the first exam into a teachable moment. Not that all of them pay attention or choose to put in the effort needed to learn, but enough clearly do that I find it to be a useful effort on my part.

No comments: