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]

Re: [Phys-l] Another uncertainties question...

Thanks Ludwik,

Even in the cases where we do have an upper limit... which was the case for many results here... I was uncertain as to how to use that in a weighted average... If it's "less than A", do you use 0, A... (something between the two?) for a weighted average? and what's the appropriate weighting?

Certainly the "less than A" probably has some basis in that for whatever given test they're using, results "fall into the noise (or somesuch)" below that level so they conclude nothing appreciable is there... But I would the question remains if that is the data I have, what is the best way to use it (ala weighted average or somesuch) to get a "best average value" of some given contaminant? So I'm not certain if that entirely answers the question for me.

Say for example I had

5.7 +- 0.5 ppm
6.0 +- 0.6 ppm
and 2 results
"Less than 5 ppm"

What do I do with that? There seems a fair likelihood that we got something like "4.7 ppms" buried in there under the guise of a <5ppm. I'll grant that perhaps the data would be better if reported differently, but if that's what you got, where do you go with it? (And in the world of industry "repeat the experiment" ain't always an option -- It wasn't in this case, so I can't give her that as an answer, either -- Even if scientifically best... ;) Is there anything that's "most intelligent" or at least "accepted practice"?

J



Ludwik Kowalski wrote:
The proper way to report the "non-detect" would be to specify the upper limit, for example, "less than A." The upper limit should ideally also be specified in terms of the mean value and standard deviation. But this is often not easy. In practice the A is set up more or less arbitrarily, I suspect.
Ludwik

On Jan 25, 2008, at 12:36 PM, Jason Alferness wrote:

(Sorry for a cross-post...)

So -- awhile ago, I ran into another problem regarding uncertainties.

For some groundwater contamination, a friend had to compute a weighted
average of some values over times and locations. Some values were given
as standard (value ± uncertainty) or equivalent, which is of course
straightforward enough. But the question I wasn't sure about was how to
deal with the non-detect results... Often they were reported as
something like: "Less than xxx ppm" or somesuch...

Is there a standard accepted way to deal with non-detects? It seems not
quite correct to do any of the following: Ignore it, call it zero, use
the threshold detect value, or to use the threshold detect as the
uncertainty for either.

Anybody had a similar experience or know what accepted procedure is?

Jason Alferness

--
-------------------------------------------------------------------------------------------
A person, who is nice to you, but rude to the waiter/waitress, is not a
nice person. (This is very important. Pay attention - It never fails.)
Dave Barry from "FIFTEEN THINGS THAT IT TOOK ME OVER 50 YEARS TO LEARN"

Jason Alferness
University of Washington
Room B256B, Physics & Astronomy Building
Campus Box 351560
Seattle, Washington.&nbsp; 98195-1560


Phone: (206) 221-2974
FAX : (206) 685-0635
email: alf@phys.washington.edu

_______________________________________________
Forum for Physics Educators
Phys-l@carnot.physics.buffalo.edu
https://carnot.physics.buffalo.edu/mailman/listinfo/phys-l


_______________________________________________________
Ludwik Kowalski, a retired physicist
5 Horizon Road, apt. 2702, Fort Lee, NJ, 07024, USA
Also an amateur journalist at http://csam.montclair.edu/~kowalski/cf/



--

-------------------------------------------------------------------------------------------
A person, who is nice to you, but rude to the waiter/waitress, is not a nice person. (This is very important. Pay attention - It never fails.) Dave Barry from "FIFTEEN THINGS THAT IT TOOK ME OVER 50 YEARS TO LEARN"

Jason Alferness
University of Washington
Room B256B, Physics & Astronomy Building
Campus Box 351560
Seattle, Washington.&nbsp; 98195-1560


Phone: (206) 221-2974
FAX : (206) 685-0635
email: alf@phys.washington.edu