Re: Media Testing for maclife.de
Re: Media Testing for maclife.de
- Subject: Re: Media Testing for maclife.de
- From: David Scharf <email@hidden>
- Date: Wed, 17 Sep 2008 19:53:47 -0700
Chris Cox wrote:
On 9/16/08 8:11 PM, "David Scharf" <email@hidden> wrote:
The measurement is not wrong - it is just a number or set of
numbers.
But the meaning you assign to the measurement would relate to
the accuracy
and precision of the instrument used to make the measurement
plus the
experimental setup from which the measurement was made.
The interpretation of the measurement may not match a
standard, or other
measurements -- but that simply means that something is wrong
with your
interpretation of the measurement(s).
The concept is incorrect. Suppose you take 2 Kiethley 2001 DVMs
and measure a NIST traceable standard of 5.000000 VDC. Instrument
#1 measures 5.0001 VDC and instrument #2 measures 4.6875 VDC. I
say instrument #2 is just plain wrong and instrument #1 is well
within specified tolerance of accuracy. According to your theory,
instrument #1 is correct, and if I understand you correctly,
instrument #2 is also correct, but with an explanation and
interpretation of the circumstance. That's a pretty far out
philosophy for an exacting science. We have calibrated instruments
so we don't have to explain the entire circumstances for all
things with every measurement. As I said, I understand your basic
point, that the circumstance is important to the interpretation of
a measurement, but the way you stated it is far too broad.
One of the instruments does not match standards -- which means you
don't know the correct scaling for the instrument. Both measurements
are valid (barring someone reading you random numbers). You just
don't know how to interpret them to match your standard. And you
define "correct" as meaning "matches the standard". If you find the
mapping from instrument 2 to the standard -- you can still use the
measurements. Instrument 1 just had it's mapping done ahead of time
(which you can't do with all measurement devices, unfortunately).
"Don't know the correct scaling"!!?? What? The premise is that we did
know the correct scaling. By "scaling", I mean calibration, perhaps you
mean mapping of the instrument's mis-calibration. In the practice of
making precision measurements, scientists and engineers rely on properly
calibrated equipment as a "first line of defense". We do not make
software "profiles" of poorly calibrated instruments in order to make
them usable. We just calibrate them against known precision standards.
It was interesting to read the post about imprecision of
spectrophotometers yesterday.
Again, the measurement itself is just a number (or set of numbers) -
by itself, it cannot be wrong.
Well, yes it can be wrong! Actually if you take the opposite attitude to
yours, one could say that most almost measurements are incorrect
(wrong). Its just a matter of how incorrect (imprecise). Although I
originally thought you had just mis-spoke, apparently you can be wrong
too. I am surprised at your attitude about it.
It is an important point in sciences: just because the measurement
doesn't match what you expected, that does not make the measurement
wrong. Usually it just means that there is something you failed to
account for in the theory that predicted the values. If we threw away
all results that didn't match expectations, much of our scientific
discovery would never happen.
Chris
Well Chris, its good that you work for a software company and don't need
to do lots of test and measurement work. I suppose that everything can
be corrected in software now, eh? 8-)
All the best,
David
*DAVID SCHARF PHOTOGRAPHY*
Scanning Electron Microscopy
Phone 323-666-8657
Los Angeles, CA 90039
http://www.electronmicro.com
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden