Brian Melton
March 12, 2011
My article this week is “Student Centered Instruction in a Theoretical Statistics Course” and is found at http://www.amstat.org/publications/jse/v17n3/batesprins.html. In teaching an introductory statistics course, I am very interested in developing a student centered approach. Unfortunately, the tendency in an introductory course is more instructor centered due to the fact that many are presented in a lecture format where there are a large number of students, personal interaction with students is difficult unless the student takes the initiative to approach the instructor, and even the classroom, such as a lecture hall, is not as conducive to a student learning approach.
In this article the author conducts her own test by first conducting courses in the first semester using a strictly teacher focused format. In the second semester, she taught a follow-on course with an alternating approach to teaching. On Tuesday she would be more centered on a teacher learning approach with primarily lecture to cover the required material. When the class resumed on Thursday, she adjusted to a student centered approach, where students worked and discussed problems. She would provide minimal guidance to keep them on track, but for the most part allowed the students to experiment, make mistakes, and learn from those mistakes. Throughout the semester she would make minor adjustment to her approach when it appeared the students were not engaged at the desired level. For example, on Tuesday, she would give one problem to turn in on Thursday, and another to be worked by the group on Thursday. When she realized the students would focus on the assigned turn-in problem, and wait until Thursday to look at the other, she assigned both problems without identifying which one was which. This forced the students to work both prior to class.
Ultimately the author found that student centered learning was much more time consuming in preparing and thinking through the lesson, however, the class was typically more enjoyable for both the students and the instructor. The author pointed out that overall grades rose by 2.5% between the two courses, and cited this as evidence that the students had improved learning.
The article was interesting, though I believe the author did not go far enough, and there is a still an open question about how well it would work in an introductory course. The two courses were actually 400-level level courses, and the class size was only eight students the first semester, of which only four took the second course. But the author felt the concepts could be applied to introductory courses. Additionally, with such a small class, I would have preferred to see the teacher use a student approach for the entire course, not just on Thursdays. With the small number of students, I have doubts about whether a mere 2.5% improvement is really being driven by the student learning approach.
Although I found the idea that we can find ways to teach statistics with a student focused approach, I am still a little skeptical of how well this author’s style could be used in an introductory course. There were some promising aspects of it from the anecdotal evidence based upon the written feedback the author received from the students, where they felt they were more challenged using the student centered approach. Still overall, I found the author’s idea that associating this particular teaching experience with what we would find in most introductory level statistics course a bit of a stretch.
Brian, I think that an instructor could teach some information at the beginning of each class and then have the class break up into small groups or pairs to apply the information to some real world problems. Statistics is very “dry” for some of us, and working problems that relate to people make the work more interesting. Similar idea is mentioned in Michael’s post about teaching math, and also in my reply.
Brian,
I did not refer to the source but from your account I share your skepticism regarding “how team learning (or student-centered) would work in an introductory statistics course” and the “statistical significance” of the modest improvement observed.
It seems like a reasonable approach in a first experiment, to let students collaborate and solve problems on their own on the second class of the week after first guiding them to understand some relevant principles, concepts, or applications in a lecture during the first class. My experience in working statistics homework alone, after heavy doses of lectures of course, both in the undergraduate and graduate schools, is that is where the real learning took place. I think it might have been more productive for me to do the homework with other students and wonder if that is a better model. Perhaps the author’s approach is more a way to ensure that every student in the class had at least one such group-work experience a week.
My other experience in learning statistics in the work place (in a Six Sigma Black Belt program) was mostly focused on exercises requiring team participation and involving role playing among the members. Those classes were engaging, but I agreed with the author of your selected paper, it was more challenging for the instructor to prepare and work out all the bugs (e.g. misinterpretation of instructions) in that format versus a typical lecture format. However, in that format, where the solution approaches were pretty much well known and mechanically applied, the deep learning seemed to take a back seat due to the environment, especially the team dynamics. My intuition is that in an into Statistics class, real learning can benefit some from the “n-heads are better than one” effect, mostly if the individuals minds are already well prepared by prior instruction and self-work. But that sounds a lot like what your author implemented in the Tuesday-Thursday plan, so maybe the 2.5% simply reflects that the dominant factor is student ability, not instructional method. Good luck with your attempt, Brian!