|
Dr. Doug
Falls, Biology, Emory worked on materials for two courses. One,
a neurodevelopment course, Building Brains (see detailed report)
and the other a new set of ideas for teaching statistical reasoning.
Doug investigated tools and approaches that might used to effectively
teach central concepts of statistics to students taking Biology
courses. He focused on the core ideas of sampling, statistical hypothesis
testing, and experimental design.
Since these lessons are intended as supplements to
courses in (for example) cancer biology and neurodevelopment, the
concepts must be learned in a few hours, mostly outside of class.
The tools and approaches to the development of these "lessons" rely
heavily on materials made available on the WWW. Simulations and
case-based studies will play a critical role in motivating study
and effectively giving students a "feel" for the "reality" of sampling
variation and statistical methods for dealing with this reality.
He has drafted a case study and surveyed many WWW sites dealing
with statistics, and has begun to prepare an annotated list of these.
He has found computer simulations that can be employed in the case
studies This summer Drs. Marsteller, Falls, Easterling and Quiñones
plan to work together on several research methods and statistics
developments.
Doug's initial Case study / scenario:
The Center for Homeopathic studies reported what they describe
as conclusive evidence that milk of dandelion is more effective
than diuretic drugs in treating hypertension. Maggie was just beginning
to plan the dayÕs experiments when the phone rang. Maggie picked
up the receiver and was surprised to hear JohnÕs voice. Though they
were very good friends, John rarely called Maggie at work early
in the morning. John was quite excited: "Maggie", he said, "I just
saw an infomercial on TV about the latest advances in medicine.
It was put together by a Dr. Carl who has discovered that milk of
dandelion is a miracle drug. ItÕs great for baldness, but even better
it cures high blood pressure. He said that they compared milk of
dandelion to standard medical treatments like water pills, and the
milk of dandelion was much better. He even had on the show several
people who had stopped taking water pills and switched to the milk
of dandelion. They said they felt much better and that their blood
pressure was great. Maggie set her pipette man down and started
to ask John about the evidence presented.
"What is meant by the 'margin of error' in a
political poll or the 'confidence interval' of a sample? What does
the statement that a scientific study comparing treatments A and
B for a serious medical condition has shown a (statistically) significant
difference between the treatments? Or that a study has failed to
confirm a previously reported significant difference? The world
population as a whole has a very shallow understanding of the basic
and critically important concepts underlying these statements. Even
most advanced undergraduate Emory students and even graduate students
in the biological sciences don't have a good 'feel' for the underlying
concepts. This includes many students who have taken a course in
probability and statistics and students who can state the definition
of 'statistically significant', but who, in fact, don't have a feel
for what this means. For most, 'sampling error' is a failure of
experimental design and statistical significance equates with 'truth'."
|
