The results on my Linear Relations exam freaked me out a bit, so I've swung back to traditional teaching: show some definitions, do some sample problems, give them time to practice those problems.
As predicted, the thinking stopped. Well not totally; It stopped during the teaching portion. When I ask students to answer questions, I'm met with a bunch of blank stares. However when given time to practice individually, the thinking appears to be there.
Today I did a combination--traditional teaching for some basics (definitions of precision and uncertainty), and group work at the whiteboards to investigate adding and subtracting measurements that have uncertainties. The group work went well, and they seem to understand why you add uncertainties regardless of whether you're adding or subtracting the measurements.
Balance. That's what I need to work on. Balance traditional teaching (for things like definitions and "basic" work), group work (where understanding of some concept is required), and individual work (so they can practice, and individual abilities don't get masked by the ability of groups as a whole).