[Phys-L] Re: Human Error?

[Hugh Haskell <hhaskell@MINDSPRING.COM>]

At 12:13 -0400 4/29/05, Edmiston, Mike wrote:
> How do you feel about students listing "human error" as one of the
> possibilities for why their lab results are not as good as hoped for?
I personally think that any student using that tired phrase should be
lined up against the wall and shot, or a least thoroughly flayed with
a cat-o-nine-tails.

I cannot think of a more meaningless phrase to use in the context of
experimental uncertainty. Students have just learned that any time
they find the results not in accord with what they think they ought
to be can cover their butts with "human error," and we should not
tolerate it. I would much rather hear them say that they cannot
figure out why their results aren't what they thought they should
have been than to try to snivel out with "human error."

Every measurement has "human error" because every measurement was
either made by a human or the apparatus making the measurement we
designed by a human. If they can't figure out that the finite beam
width of a photogate, or the response time of a temperature probe, or
the parallax of the meter needle, or the stretch of a sting used to
measure a length, or whatever, can contribute something to the
imprecision of every experiment, then they don't deserve to pass the
course. And if they cannot give at least a rough estimate of how much
whatever measuring device they used contributed to the imprecision of
the result, then they simply aren't doing what they should be in lab.

Another of my pet peeves is their habit (usually learned in other
science courses) of taking the difference between their result and
the "right answer" and calling that the "experimental error." I
always ask them if that is true, then how does someone who makes a
measurement for the first time determine what the "right answer" is,
so they can figure out their experimental error.

If the experiment is such that a predicted answer is appropriate,
then when they do their measurement, I insist that they provide some
"error bars" (with justification) around their result, and they may
not conclude that their results {agree with current theory" unless
the currently accepted value of the thing they measured lies inside
the error bars of their result. I then tell them that if it lies
outside the error bars, then they either need to provide me with a
new theory to account for the difference, or redo the experiment
until their error bars contain the accepted value.

To avoid the problem coming up, I also design as many labs as
possible for which the numerical answer they seek is not some
standard quantity--that is, the number they seek is dependent upon
the setup they use, so every student's results should be different.

i sympathize with your problem, Mike. The only way I ever beat that
problem was to simply not accept papers that contained the phrase. If
they wrote that, I returned the paper and made them redo it.
Unfortunately, there are a whole lot of teachers (mostly, I think, at
the middle-school level) who really do think that is an acceptable
form of error analysis.

Hugh
--

Hugh Haskell
<mailto:haskell@ncssm.edu>
<mailto:hhaskell@mindspring.com>

(919) 467-7610

Never ask someone what computer they use. If they use a Mac, they
will tell you. If not, why embarrass them?
                        --Douglas Adams
******************************************************
_______________________________________________
Phys-L mailing list
Phys-L@electron.physics.buffalo.edu
https://www.physics.buffalo.edu/mailman/listinfo/phys-l