In the past, I’ve talked a lot about why I love certain kinds of misconceptions. In this particular post, I talked about why I love the misconception that the earth gets closer to the sun in the summer. Two recurring claim of mine have been that (1) student ideas should be evaluated with respect to the evidence and reasoning they currently have available, and that (2) sensible, explanatory, and well-articulated misconceptions are to be cherished over impoverished but accurate scientific statements.
In this vein, If you have never read Philip Sadler’s 1998 article, “Psychometric Models of Student Conceptions in Science: Reconciling Qualitative Studies and Distractor-Driven Assessment Instruments” in the Journal of Research in Science Teaching, you might want to.
To steal a quote from its abstract:
…instruction appears to strengthen support for alternative conceptions before this preference eventually declines. This lends support to the view that such ideas may actually be markers of progress toward scientific understanding and are not impediments to learning.
Misconceptions are markers of progress. Yes, he said it.
To give you yet another reason to read the paper. Here’s Figure 2 from page 276, showing the popularity of different ideas to explain the seasons at different “ability levels” collapsed along single dimension. What patterns do you see?
It is true that often times when we see misconceptions in class, we gasp. But here we have yet another reason to think about misconceptions as important for learning and possibly necessary to make progress. This year, I’m going to brag a lot more about all the misconceptions that come up in and because of my class. I’m going to brag because it might mean that my students are making more progress by developing misconceptions than by either idly sitting around not thinking about the world or by trying to memorize correct scientific statements. Be a good teacher this year: go out and cause some misconceptions.
Teach. Brian. Teach. 
You write:
I’m curious what patterns you see; other than that only the high-ability students hold the correct conception (and I haven’t read the article, so I have no idea what high ability means.
Another way of justifying your preference for celebrating student misconceptions is that these misconceptions are ideas. When we bring out student misconceptions, we are examining their ideas. And if I can get their ideas out there, I can help them make those ideas better. If the misconceptions are hiding, though? They’re still there. I can pretend they are not, but they certainly are. And I can’t help to change an idea that I cannot, or will not, see.
The pattern I see is this. Let’s say you are doing inquiry into astronomy topics and on an assessment piece you find that 30% of your students have firm notion that earth is closer to sun in the summer and the rest seem to have a random collection of ideas that are neither here nor there. You engage them in thinking over a period of weeks about astronomy topics and find that later 60% of your students are now more committed to this wrong idea. Intuitively, one might think, “Oh my gosh! I caused more of my students to be more committed to this wrong idea”… But the graph is showing us that by doing so you’ve actually nudged them along in their understanding. By thinking and talking about their ideas, even wrong ones, they are now closer to expert thinking. It is for exactly the reason you say: it’s not that you are encouraging misconceptions, but that you are encouraging ideas, and by doing so helping those ideas to make it to the level of public discourse.
This pattern makes me pause and rethink how we assess. How often would a teacher see a classroom full of more students committed more strongly to wrong ideas as a sign of progress? How often would a standardized exam look for the prevalence of specific wrong ideas as indicators that students have begun a journey to understand complex phenomena? The suggestion is perhaps that progress through (science) learning is not incremental steps toward correctness, and that an insistence on correctness at small time scales is counterproductive for learning.