Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: Feynman (was pedagogy)



Well, what evidence is there that modeling mistakes produces better problem
solving skills? Feynman certainly had little evidence for his methods, and
he called his lectures failures.

Now there is evidence that improving conceptual understanding also improves
problem solving skills somewhat. Eric Mazur has cited such evidence in his
book "Peer Instruction". The UMPERG group uses concept based problem
solving. Alan Van Heuvelin has developed ALPS worksheets which help
students apply concepts to problem solving. The work of the Heller's does
show improved problem solving skills, and as far as I know the instructor
seldom shows students how to work specific problems.

OTOH the Modeling program does not do any "modeling" of problem solving, but
rather has the students exchange all of the ideas. There is some very good
evidence that this approach to problem solving is quite superior to having
the instructor model the problem solving. It is interesting that Modeling
uses some very conventional appearing labs and problem sets, yet achieves
fairly high FCI gain. I think that Modeling could benefit from improving
the problems by making them richer in context. Indeed there is evidence
that the traditional method of having the TA solve problems actually
decreases the ability of students to solve problems.

One of the most dramatic increases in problem solving skills probably is the
work of Miehl at Capetown where he turned a 50% failure rate into a 100%
passing rate. This was done by some very careful research and the students'
cognitive problems were specifically addressed.

Remember that logical, plausible, and attractive can be seductive traps.
The Impetus theory seemed that way until Newton came up with a better
alternative. Similarly quantum theory was at first very illogical until we
found ways to understand it. We need evidence for the particular methods,
and then we need to use methods that produce dramatic changes, not just
small gains.

Priscilla Laws who has been in the business for a long time said that her
intuition for what works has improved, but that she is still surprised when
perfectly reasonable, "plausible & attractive" ideas produce the opposite
from the intended effect.

In other words my beef is that solving problems in front of the class has
limited effectiveness at best. And doing this frequently is likely to
degrade the thinking of the students. Students want teachers to do this so
they can memorize problem solutions. Just because both teachers and
students have bought into this does not make it very effective. As long as
the idea is not challenged, most instructors will continue to think it is
effective.

I will also admit that we do not have good standard evaluations for problem
solving the way we have the FCI, FMCE, ... and other evaluations by various
researchers of conceptual understanding. But we should be trying to look
for evidence. For example has anyone tried to model problem solving with
and without the "mistakes", and then looked for evidence in exams?

John M. Clement
Houston, TX

PS: My last name was originally a first name. My great-grandfather was
forbidden to use the family name when he became an actor so he used his
first name and it has no s.


Thank Goodness for John Clements, who reminds us time and again,
that no matter how plausible & attractive the teaching initiatives -
(e.g. to demonstrate problem solving cold - complete with warts) -
if there is no experimental improvement shown, then there is no
scientific basis for a method.

Brian Whatcott