I just read a paper about Open-ended Learning Environments [Land, S. M., & Hannafin, M. J. (1997). Patterns of understanding with open-ended learning environments: A qualitative study. Educational Technology Research and Development, 45(2), 47-73.] which discussed using software to design a roller coaster, in addition to some resources to learn about some of the principles involved to help them design one that doesn't crash. It was open, in that students could play with the simulator and just see what happened and were free to use or not use the other resources to help them out. It was a case study, discussing the experiences of the four students involved, to see how much they would learn without being directly taught or guided by an instructor. I haven't read as many qualitative studies as I have quantitative studies, but this one seemed a bit more organized and well-planned than some qualitative studies I've seen. There was definitely a methodical, scientific approach in everything, from choosing the participants to the method of coding their actions and responses, while retaining a case study approach.
The two main items discussed in the paper were the differences in how students applied prior knowledge while forming either informal or formal hypotheses and the extent to which students followed a methodical approach in testing those hypotheses. There was little interaction with the learning materials that were available, with one student confused and frustrated with the lack of an "answer" in the materials as to what combination of horsepower, weight, hill height, and turning radius was "best".
Each student had different ideas of what was successful. For one, the fact that the coaster did not crash meant that it was successful. For another, if the coaster was as close to crashing as possible, but didn't, that was successful. There was not a clear explanation of how students performed on the quizzes, other than one student who answered a question correctly, but in practice actually acted in a manner that would suggest a lack of understanding of that concept. It would be interesting to compare the performance of these students to others that were taught the same concepts in a traditional classroom environment, but that is not done in this paper.
One interesting quote from the conclusion of the paper was that "the effectiveness of the constructivist environment relies heavily on the learner's task management and decision-making processes." Of course, this is somewhat of a loaded observation. If the effectiveness of constructivist learning environments is in question, and we are immersing students who have been trained in a dichotomous student/teacher banking environment (teachers depositing knowledge into student brains), as challenged by Freire and Dewey, it is hardly surprising that students may be lacking in some of the skills that would allow them to keep themselves on task and make effective decisions.
A few times when I have come in as a new Scout leader for a group of boys that have been oppressed and controlled by the previous untrained Scout leader, and I begin to allow the boys to make their own decisions (since they are supposed to run their own program with ideally minimal guidance from their leaders), chaos ensues. I assume this is the case, because when given a little bit of wiggle room, which they have never experienced before, they go out of control because they don't know how to handle the freedom. It takes some time to reign them back in, and eventually they learn to cope with freedom, which may be more difficult to deal with than oppression. I also see similar issues in my work with CIL here on campus. There are many ways for students to prepare for the CIL tests, whether coming to formal classes, attending informal demonstrations, reading on their own, taking practice tests, or just trying the real tests. There is no set "best" way to study and no particular deadline that all students must meet (although individually there are deadlines, such as applying to a program, registering for a class, completing graduation paperwork, etc.). We often see students that follow a less than methodical approach of simply taking a test without preparing, and after they fail it, they immediately take the test again, apparently hoping they'll guess just differently enough to pass the test.
So do constructivist learning environments fail because of inherent problems in the process or because we have trained students (and been trained ourselves) that the teacher's job is to stand in front of a classroom expounding and exhorting while the student's role is to sit there and bask in our glory, soaking up the knowledge that we spew forth? Obviously that statement is a bit over the top and is not meant to say that I believe constructivism is the answer to everything, rather that it is difficult to create a study that can untrain years of practice by both teachers and students in a traditional pedagogical environment.