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]

[Phys-L] Re: Human Error - not tolerable?



John Denker criticised the last paragraph (of my previous post) which
began "In the end, human errors are not tolerable."

I think John's criticism was very good, so I am looking for new words to
convey what I really want to convey.

I am an educator with many students who make foolish blunders. (1) Some
students turn in reports they know are wrong, but they just don't care
to fix them. (2) Some would have fixed the blunder if they had more
time, but they only discovered the problem the night before the report
was due. (3) Some never realized the blunder because they didn't have
any idea what to expect; i.e. they didn't think enough about the
experiment to realize there is no way the 1-kohm 0.25-W resistor on the
breadboard could have a 100-amp current through it.

I want to convey that in my class problems (1) don't care, and (2) too
late, are not tolerable. Students who don't care about lab results
should not aspire to be scientists. Even if they think of themselves as
theoreticians, they ought to care enough about the foundation provided
by undergraduate labs to do the labs correctly. In general, students
who don't care about their school work should not be paying tuition in
the first place.

Although John is correct that some students are in a situation where it
is "too late" once they leave their assigned lab period, that is not the
case in my classes. I am fortunate. I can do the spiral approach John
described. I can attempt to give the students a taste of typical
scientific procedure... take some data, analyze it, take some more data,
analyize it, refine the experiment, take some more data, etc. For this
privelege my students (or their parents) are paying at least twice as
much tuition as they would pay at a state-supported college.

[Aside... Although we do not accomplish 14 or 15 separate experiments in
one semester, it is not as bad as N/2 like John suggested. My
assignments overlap. My goal is 8 to 12 experiments per semester. It
depends on the class, and 10 is typical. Students get a new lab
assignment about every 1.5 weeks. The allotted time from assignment to
report is usually 2 weeks and occasionally 3 weeks. That means they are
often working on two labs, and occasionally three labs, at the same
time. The idea is that they are beginning to take data for a new lab
while they are also analyzing and writing up the previous lab. If they
discover the need for more, different, or better data for the previous
lab, they can get that data on their same trip to lab for the new
experiment, or they can schedule a separate time. Typical data
acquisition for my experiments only takes 30 to 60 minutes. I do not
make them come to lab for a 2-hour or 3-hour block, although some will
spend that much time. I think it is good for them to have multiple
irons in the fire.]

[My course is a 5-hr course. I think I can expect 15 hours a week from
them. We meet all 5 days of the week, so I can expect 10 more hours
outside of class. The good, organized, concientous students tell me
they can get an A or B grade and only spend 5 to 8 hours a week outside
of class, so I don't think I am being an ogre.]

Let's return and look at point (3). I have more sympathy for students
who don't have a feel for the kind of data to expect. This is true for
all kinds of experiments, but especially true for electricity
experiments. Students have no feel for voltage, current, resistance,
power, etc. However students need to learn they are not alone. In
college they have fellow students, and in particular they have a
professor. Once they are out of college, these resource people are
called "colleagues." When I was in graduate school I often went to
other grad students, or post docs, or my committee members, and asked...
"do these results make any sense?" Likewise, when I worked at Los
Alamos, I would often ask other people in my group what they thought.
Indeed, we had weekly group meetings where we reported on our current
problems and successes. Even if the discussions tended toward
kibitzing, it was generally valuable and appreciated.

When students are unsure of their results, yet they fail to ask me or
fellow students if their data or results make sense, what is the
problem? Too shy? I can buy too shy (or embarrassed) and I work to
bring down those barriers. Too lazy? I cannot allow too lazy.

But... John is still right. My statement that "human error is not
tolerable" is too strong. What if I expand the last paragraph of my
previous post into the following two paragraphs.

In the end, "human error" is not a phrase found in error-analysis
sections of scientific papers. That doesn't mean human errors don't
occur... it just means readers assume you took reasonable steps to
discover and correct human errors before you published. The fact that
undiscovered human error might still exist is understood, and you should
not explicitly say so. If you have not taken reasonable steps to find
and fix human errors, you are probably not ready to publish. However,
time and expense often modify the meaning of "reasonable steps." You
may have little time (before a deadline) to discover blunders. Some
experiments are very expensive, and redoing them to help assure data
consistency and to expose human errors might not be affordable even if
you have the time.

Perhaps you are aware of a blunder, but time or expense prohibit fixing
the blunder. Even so, you might need or want to publish what you have.
For example, your results might still be useful to the scientific
community, blunder and all. Perhaps you must report your current
experimental status in an organizational annual report, or in an annual
report to funding agencies. However, if you have reason to publish data
with known human errors, you can't simply state there was human error.
You have to describe the specific error and discuss the implications or
consequences.

Of course, if telling students that human error is not tolerable does
not prevent them from using that excuse, the above paragraphs also will
not work.

Michael D. Edmiston, Ph.D.
Professor of Physics and Chemistry
Bluffton University
Bluffton, OH 45817
(419)-358-3270
edmiston@bluffton.edu
_______________________________________________
Phys-L mailing list
Phys-L@electron.physics.buffalo.edu
https://www.physics.buffalo.edu/mailman/listinfo/phys-l