Re: what good is "percentage error"?

[Doug Craigen <dcc@cyberspc.mb.ca>]

On Wed, 14 May 1997, Rauber, Joel              Phys wrote:

> You're right we were talking apples and oranges.  I basically agree with the 
> above.  Although, I do think the student should try to get the "right" 
> answer even when known in advance; WITHIN THE LIMITS OF THE EQUIPMENT.  And 

I suppose that my approach goes something like this, I expect a students
lab report to include the following:
1. find the answer
2. estimate the accuracy of the answer
3. compare the answer with the expected answer, or provide other 
   "sanity checks" where an exact prediction is not available
4. discuss #3

In general, small discrepancies in #3 and reasonable values for #2 are
necessary for a good mark and make #4 much easier for the student to
write. However, I will not make presumptions about 1-3 until I've read #4
when I'm marking the report.  

In fact small systematic errors in equipment may be seen as a sort of
blessing.  Students developing good experimental skills will find the
errors in step #4, while other students will get "suspiciously accurate"
answers. 

I suppose there are two valid objectives that conflict a bit here.
One is teaching students to make the best and most accurate use of a known
piece of equipment for obtaining a measurement.  The other is teaching
about experimental design, which necessarily means making mistakes and
tracking them down, finding out for yourself the sytematic problems in
equipment ... Personally, I like to see evidence of the latter in a
student's report, however there is not always time to then go back and do
it better based on what was learned.