The Number

Just looking over my teaching evals from last fall. Mostly good stuff, but there's always the occasional sour note. One student complains I don't seem to be teaching the unit on scientific thinking in accord with Christian principles. A few kvetch about the amount of reading required for a senior capstone. By and large, though, it's 90-95 percent on the positive side. But here's the thing: these evals are fragile at best and almost meaningless at worst.

I always have to remind myself how unreliable the evaluation numbers actually are. How do I know they are unreliable? Well, consider this: last semester I taught two sections of a senior capstone for the general education core, one at 8:00 am on T-Th and another at 9:30 on the same days. Same class, same material, same approach, same assignments, same everything. Yet the scores are very different.

Now in my 8:00 am section the scores were generally in the upper half of the middle 40 percent, which means I'm about where most people are across the country, maybe slightly to the right of the bell curve's center line (mid 50s to low 60s). Okay, that class files out and 10 minutes later a new group files in. Now my scores jump to the top of the second highest quintile (upper 70s to mid 80s). Again, same course, same day, same material, same approach....

The only variables are the time of day and the students. Yet were I untenured, I'm the one who would be held accountable to these numbers. Well, to be honest, the numbers are good overall, so I probably wouldn't suffer any adverse consequences. But the point is that something as simple as scheduling a class at 8:00 am affects the scores. You should see what happens to the numbers when you teach a late evening course (8-10:25 pm). Keep in mind that such a course is mostly populated by people who have to go to work in the morning.

But we must have a number! You would be amazed at how much these numbers are relied upon by various people. These are smart people, too, many of whom have been trained in statistics and measurement; but give them a number and that's good enough. It's almost like they were stock analysts looking at Value at Risk data and models. All those quant jocks on Wall Street had plenty of numbers.

Didn't really help, did it?


Popular posts from this blog

Two Jars

The Betrayal of F. Scott Fitzgerald's Adverbs

Four Arguments for the Elimination of the Liberal Arts