How computing and physics learning differ
Allison Elliott Tew successfully defended her thesis proposal this morning. Hooray! You may recall from my pre-SIGCSE description of her work, that she’s attempting to build a language independent measure of CS1 learning. Allison talked about concept inventories in a way this morning that I found intriguing with respect to computing.
You may know that concept inventories are used to assess student knowledge about an area. They are based on an analysis of what students already think about an area, and those pre-conceptions/misconceptions appear as “distractors” in the multiple-choice questions. The most famous of these assessments is probably the Force Concept Inventory, which was developed over 25 years by David Hestenes. The FCI measures knowledge about Newtonian mechanics, and it includes all those deeply-held beliefs that students have about the world from living in it for 18 years before entering a College physics classroom. The FCI was used in a huge study (n>6000 students) by Richard Hake to show that instruction alone was ineffective in shaking those beliefs, and “interactive engagement” (like peer instruction) was necessary to get students to learn physics well.
There are efforts to build concept inventories for computer science, but they run into a problem when creating a direct mapping. Students enter our classes, for the most part, without any conceptions at all of computing. If they have conceptions (or misconceptions), they’ve only developed them recently. Students may have an idea about how computers work from years of working with those computers, but the challenging issues of variable types, defining classes, pointers, recursion, and essentially everything that students have trouble with in computing are all totally new to their computer science classes. They don’t have 18 years to develop preconceptions and misconceptions. That makes it hard to develop a CS concept inventory, because any wrongly-held beliefs that students have are due to their instruction, not due to naive reflection on experience.
Which leads me to my question: How deeply held are those misconceptions? Physics misconceptions are well-documented and very hard to shake. They have served 18 year olds very well! How about computer science misconceptions? Since they form over a short period of time, can we just correct them with, “No, that’s wrong”? Maybe if we taught things better, there would be no misconceptions to inventory. And if there are some, maybe they’re really easy to change. I don’t know how one would measure strength of misconception, but I’ll bet that it’s different between physics and computer science.