I thought I should point out that my decisions about how to organize TBL
in my class have been influenced by some practical considerations. One
is that introductory statistics texts for graduate students usually have
units or chapters that take no more than one week. This suggests a RAT
every week that there is no exam. Another is that I started using TBL
when my classes were scheduled for three-hour sessions once per week. (I
now have two two-hour sessions per week.) As a result, I had a RAT or
exam every session.
While I would prefer that students come prepared with long lists of
questions about their reading and problems, this rarely happens. On the
few occasions that this has happened, because questions and answers took
quite a while, eg, 40-60 minutes, I have cancelled the IRAT but never
the TRAT. This might not work well for courses where the natural units
take more time, eg, two to three weeks apiece, since there is less
feedback from the students.
The RATs are quite useful as prologue and practice for exams. I think
they relieve stress, particularly before the first exam. With three
exams, this becomes less important after the first two.
I use almost the same RAT procedure for exams with an individual and
then a team exam. I first picked up the group exam from some of the
discussion groups in math, but the TBL approach contributed systematic
administration. At the end I give out an answer key to every student
and have the team grade the team exam before they leave the room. I
take all the exams for subsequent grading. From a behavioristic
perspective, they get immediate individual feedback from both the team
exam and the shared grading. I do very little exam review in the next
class session because it isn't needed-students have already figured out
where they stand. I usually say a few things about fairly widely
dispersed errors and this takes 5-10 minutes. The teams do extremely
well on the team exams, better than on team RATs, less than 100% is
rare.
Questions tell me a lot about what students are thinking. The students
do not seem to be aware that their questions lead to what I consider a
lecture. By this I mean that questions are not answered alone, but are
elaborated as to the bigger issues involved. The subject matter is
alien to them and I need to hear them use the words and concepts of
statistics. (Sometimes I think it might have something in common with
teaching English as a foreign language.) I can't know what students
have internalized without hearing them say things. For the same reason
I use short answer rather than multiple-choice RATs, because I need to
see students write down what they think statistics says.
I always lecture just before a new chapter, about once a week. I try to
hold it to 25 minutes, but it often creeps up to 40. I think this
bothers me more than the students since I know that they won't remember
much of the material after the first 20 minutes.
Some of my lectures are integrative, and unconnected to a single topic.
They occur after a few weeks and explain how statistics hangs together.
They aren't helpful until students have seen several kinds of tests and
confidence intervals, but scheduling them also depends on judging
whether students are ready to hear that life is easy because all
statistical tests are really variations on a theme, similarly for
confidence intervals, and confidence intervals are actually quite
closely, formally, related to tests. This completely contradicts their
views when they start the class, so I think I have to do it carefully.
The students probably don't regard my short, introductory lectures or my
question-based lectures as lectures.
Regards,
David Smith
David W. Smith, Ph.D., M.P.H.
Associate Professor, Biostatistics
The University of Texas School of Public Health
San Antonio Branch Campus
voice: (210) 562-5512
e-mail: [log in to unmask]
or [log in to unmask]
|
