Re: what good is "percentage error"?

["Michael N. Monce" <mnmon@conncoll.edu>]

On Mon, 12 May 1997, Lanzafame, Frank wrote:

> In the quantitative analysis course, students are taught to analyze samples 
> with accuracy in the parts per thousand range.  Students often analyze a sample 
> whose composition is accurately known and supplied by a company specializing in 
> supplying such.
>
> IF a given sample is capable of being analyzed to an uncertainty of 2 parts per 
> thousand, then it is expected that the student perform at that level for the A 
> grade.  The B, C, etc. grades are assigned to increasing deviations from the 
> "true" or accepted value for that sample.
>
> I'm not certain of the origin of your "horror", but the grade is designed to 
> reflect the accuracy with which the analysis is performed in a course which is 
> designed to teach quantitative methods.


	So this grade is based purely on evaluating their technique only
with the assumption that if their technique is proper they will achieve a
result to within 1% of the supplier's value..  I have several questions
about this, based on my experience in spending a year trying to undo some
of the habits these students come into physics with.  First, do you have
the students evaluate why the established technique is capable of yielding
a 1% discrepancy?  If they get greater than a 1% discrepancy, are they
then made to determine the source(s) of the increased value?  What are the
limits of the supplier's value?

	I still need to be convinced that such a grading technique is not
a pedagogical error.  The students arrive in my physics classes convinced
in the "rightness" of any value given in the book.  They are obsessed with
the "percent error".  They are shocked to learn that if they measure a
value for g to be 9 +- 1 m/s^2 that their value is actually in agreement
with the "accepted" value, as long as their uncertainty is an accurate
calculation of the limits of their measuring system.  They never seem,
especially at the beginning, to be able to question the accuracy of their
equipment, or the experimental technique being used.  All "percent error"
is attributable to "human error".

	Is the purpose of the course simply to teach established
techniques and methods which are regarded as simply skills a chemist
needs?  Then, maybe, the grading is appropriate.  But you should be aware
of the ideas this plants in the student's mind as to how experimental work
proceeds.


					Mike Monce
					Connecticut College