Thursday, May 24, 2007

An aside on "linearity"

During testimony about whether linearity was important, Dr. Brenna made the statement that "it's ensuring the yardstick measures the same at six-inches as it does at six-feet"

Which I thought illuminating-- six feet is rather off-scale for a yardstick.

12 comments:

Anonymous said...

Was waiting to find a place to plug this in. I was somewhat pleased to learn that the French are like most American males. regarding the lifting rings: apparently there is a (probably male) gene that tells us that we don't need to read the instructions when setting up a tool/device/anything with parts.
pcrosby

Anonymous said...

Almost as off the scale as the retention times...

Anonymous said...

"it's ensuring the yardstick measures the same at six-inches as it does at six-feet"

Would that make 6 inches = 3 inches? Or 36 inches = 2 yds.??

Anonymous said...

I think he was talking about linearity being accurate, repeatable and definable. The yardstick metaphor was just clumsy way of saying a ruler measures the same whether you're six inches away or six feet away.

Anonymous said...

3:16,

What I think was being pointed out a bit tongue in cheek, was that the "6 feet" distance selected could not be measured with a single yardstick...which is similar to FL's experts saying that the IRMS relative retention times were similarly off the scale.

Sachi Wilson said...

Don't forget the French use the metric system - they may not know what a yardstick is anyway. ;-)

Anonymous said...

I wonder if they know what a Dick Pound converts to in metric.

Anonymous said...

Anon 3:16

That's a mischaracterization....

Those were his words, under oath.

Its on record that he didnt know what he was talking about.

He risks perjury if he says otherwise.

Anonymous said...

The linearity issue, as I hear it in the differences between Dr's Davis and Brenna:

Davis says must verify linearity frequently as the machine is not stable and drifts in and out of linearity (normally!). Add to this the improper setup and use of the machines!!! Pointing to the correct way to verify linearity for use during the tests thus relying on the .2s, etc. This is the documented, industry standard way, that all of Landis tech witnesses seemed to expect as a prereq for good and proper work.

Brenna (and his ISL!!!) seemed to be attempting to accomodate non-linearity with there attempts to calibrate the machines performance for particular measurments to make relative retention times useful using the (in house?) method of peak patterns. I have heard nothing that suggests this is an acceptable method. It certainly seems to not the the manufacturers specified procedure! The need for this accomodation seems to suggest the systemic improper setup of the machine at LNDD.

The result of Brenna's method would then be a rubber yardstick that is only made useful by comparing a small portion to a small steel six inch ruler.

That is my guess anyway, I am not an expert.

Great JOB TBV !!! Thanks,
Russ

Anonymous said...

A few further comments on my previous post:

Davis (standard) method yields directly identifiable peaks based on the time relative to start (a linear progressive time measure).

Brenna method requires the dubious visual identification of peak patterns in both gc and irms that, according to experience are in a possibly suitably linear portion of the rubber yardstick (deemed suitable by LNDD ISL of unknown basis).

Thanks,
Russ

Anonymous said...

Essentially what he's saying is that by ensuring linearity the margin of error is reduced. I.E. if you use a 250ml flask to measure out 1 L and the margin of error on that flask is 0.5 ml, then your margin of error is 2ml. If you use a 1 L flask to do the same measurement with a margin of error of .5ml, then your margin of error is .5ml. A much more accurate measurement.

Anonymous said...

This is Russ again,

That is an reasonable explanation of one benefit of linearity from the viewpoint of accuracy but that isn't really what I am getting at. In fact, on that point. I think the Brenna method (using the machine in a broken configuration per Davis) only maybe semi-validates the 'linearity' in a portion of the machines operating range, perhaps just enough to make relative peak positions useful for identification. The misuse of the machine can only degrade from it's potential accuracy and linearity.
Linearity implies two things with these measurements when used per manufacturer.... one is that quantity measurements are repeatable and accurate to within, I think it was 1% (and that would be throughout the instruments measurement range) and .2 seconds, again through out the instruments range.
If the machine is used properly, no leaks, validated to be currently linear, then the peaks should be well separated and accurately measurable for quantity to within 1%.
The method used by Brenna seems to only arguably establish some basis for claiming, lets call it measure-ability, in a sub-portion of the machines range that are anticipated to be of interest based on the selection of mix-cal fed into the machine. The issue of difficulty of then analyzing a complex urine sample compared to 'shooting fish in a barrel' when analyzing the mix-cal which has no hopefully no surprise contaminants would seem to be more daunting under this misuse of the machine I say more daunting as Dr Davis indicated that the urine analysis was difficult with a properly operating machine.

thanks,
Russ