Try this as a starting point: http://www.cogtech.usc.edu/publications/hmelo_ep07.pdf
Teachers doing a 1 year graduate diploma don't have time to learn 20 years of euphemisms crusted onto euphemisms, just so the professor doesn't have to admit that the best methodology is "a little of everything", not the radical fads they were advocating in the 70s.
Guided instruction: "Have you thought about this? What would change if you did that?"
Direct instruction: 2+2 = 4. 2+2 = 4. 2+2=4. 2+2=4. (repeat until the student exhibits the desired response to the stimulus).
Direct instruction is indeed effective if what you're after is a set response to a given stimulus. Some would question, though, whether this is sufficient to produce an individual capable of autonomous thinking. Behaviorism is awesome if you want an animal to press a bar on cuee, less effective if what you're after is (e.g.) a person capable of evaluating the claims of politicians and making the most desirable choice (for that person) in the voting booth.
Your paper hit pretty hard on the applicability of direct instructions for simple, algorithmic tasks (e.g., the number of arithmetic problems solved per minute), precisely the type of task where memorization and regurgitation is a valid strategy.
You can't teach creativity. OK, maybe you can, but the approach of "throw the students in a group, and tell them to do something creative" is probably not exactly optimal.
Do teachers give any kind of useful guided instruction on high-level skills? Or do they just grade the assignments based on presentation?
Both approaches might have their place, but the education establishment would rather win the argument than be right. Typical arts wankers.
Here's one article. http://www.binder-riha.com/PT_DI.pdf
It took me ages to find much interesting - a google scholar search on "education" gets about 1.5 times the results as "education longitudinal", but if "direct instruction" gets about 10 times as many articles as "direct instruction longitudinal". Really, educationalists only pay lip service to empiricism. When they actually start comparing methodologies, they just use Sophistry as a methodology.