A screening tool for all diamond grading

[psadmin]

There is a new article on Pricescope journal by Dave Atlas

"A screening tool for all diamond grading "

http://journal.pricescope.com/Articles/54/1/A-screening-tool-for-all-diamond-grading.aspx

I want to thank him for his contribution to the Pricescope community.
Andrey
PS Admin

 

[strmrdr]

max confidence level of 70% seems pitifully low is that an accurate number and where did it come from?
What is the confidence level between AGS and GIA themselves? (it is less than 100% that is certain)

What is the chart based on?
Your opinion or has a study been done?
If its your opinion then it needs to be labeled as such, its not that your opinion isn't great :} it just needs to clearly say so.


Maybe do a peer review or a survey and add it too it?

 

[oldminer]

ImaGem testing of existing GIA grading played a significant role in what we know about how closely individual graders come to matching eachother. The best human color graders can do is 65% matching of grades and the rest, virtually all of the remaining 35% are one grade away. Only rarely more than a grade. One must remember that it is a bell shaped curve. 17.5% on the high or low end of each color is a borderline call. Really, this isn''t a bad statistic, but it sounds and reads as if it was an unusual problem. It isn''t. Grading to the level of color grading we now do with our eyes in the colorless end of diamonds is fairly tightly categorized. There are a lot of borderlines considering the relatively small amount of actual color difference between D and I color.

From all the evidence I''ve seen, and from what all dealers tell me, AGSL does about the same job as GIA in accuracy. They are up against the limit of highest human performance. Every once in a while, they even make a mistake or change a grade under review. Both labs are clearly honest operations.

Other labs can''t possibly do better with human grading. If they could, they would have offered it long ago.

The chart is a "SCREENING TOOL", not rocket science. It is absolutely a brainchaild of my own opinion process and not based on years of study. I want to give consumers a simple yet meaningful way to understand that there may be compromising factors in the accuracy of anyone''s or any lab''s grading report. The exact weight of each factor was definitely kept simple. I like subtraction better than logrithms(sp). Beside, I can subtract, but am worthless with a scientific calculator. If one has a .6 score, there is a sufficient warning for caution. I didn''t think it needed to be .643 or weighted in some complex manner. I am sure it could be, but I''m not the engineer to do it.

If there are further "warning" factors that a consumer should recognize, we''d all benefit from adding them in.

 

[strmrdr]

Thanks Dave that clears things up.
I thinking adding something like "Based on my years of experence this is what I find" or simular to the chart intro would be good.
I like the article and think it will be useful, thanks for taking the time too make it :}

 

[Modified Brilliant]

Dave...this is an interesting chart. Thanks for the effort.


Could you clarify this statement?

3. A grade supplied by an unknown independent lab or unrecognized grader.

"Unknown by whom?" "Unrecognized by whom?"

I may be very capable and an excellent grader but not well known outside of Massachusetts..

Jeff Averbook,GG

Graduate Gemologist/Appraiser
enjoying my 30th year in the industry.

www.metrojewelryappraisers.com

 

[oldminer]

Sure, I can explain it.

If a local client of yours, a person who knows of your excellent service and expertise uses your firm, you are "known" to them. They may safely rely upon your opinion more than someone in a distant place, or someone who has no way of ascertaining to what level of capability you grade to. For that distant client your firm, or you, are "unknown" without some knowledgeable reference that you are highly qualified. The confidence in your grade would be reduced if they don''t know what you do and how well you do it. I think it stands to reason.

So, a lab or grader who is "known" in one instance can also be a lab or a grader who is "unknown" in other circumstances. It depends on the situation of the consumer or the user of the grading report.

I hope this clarifies things a bit.

 

[oldminer]

A few suggested revisions have been made and Andrey posted the revised chart for me. Please download the newest version when you have an opportunity.