Slabs or no slabs?

[GarryN]

"No damage, just porous." Thats pretty funny. Yeah, I will give him $325 for that.

Hey speaking of holders, did anyone see the November 17 Coin World, they summarized the results of 15 coins that were submitted to about 10 grading services. PCGS was the most conservative grader overall. SEGS was next and NGC third. The other ones such as ACG, NTC, PCI and ANACS were all on the liberal end. One coin was graded AU58 by PCGS and was graded MS65 by another service. A SLQuarter was graded AU58 by PCGS and MS64 by others. Very strange. More than strange was that they only summarized 11 of the 15 coins tested and none were copper. NONE. There is also some data in a May? edition of Coin World. Maybe one of the other 4 coins was a copper. Sorry, I dont have the story in front of me right now.

[smed]

Remember... the author himself said more than once that the 15 coin sample is too small to yield necessarily valid results.

Does ANYBODY honestly feel that PCGS is the most conservative and Crap-U-Grade the most accurate? I sure don't.

One nice thing about the article, was that he BLASTED the PNG/ICTA survey as crap... which most of us probably knew that it was.

[GarryN]

I agree 15 coins are too small a sample. 100 coins would be quite an enterprise. I would like to see more than that, with each denomination and each type covered. But that would be very time consuming and expensive and it probably wouldnt stay a secret for long either

[smed]

They're all probably on the lookout for large varied lots. It would probably be better to do 10 lots of 10 using different names/addresses. Wish I had the funds to do that...

[Robert]

(connecting the dots)

It seems that there may be several posters here willing to participate in another study?

[smed]

That would run into THOUSANDS of dollars.

Let's do a ballpark guesstimate. Figure 10 bucks a coin on average. Probably about 30 for postage (15 each way) for each 10 coin lot.

What are there, eight companies tested?

Each 10 coin lot: $130 grading/shipping

$1300 for all 100 coins PER service

8 times $1300

(drumroll please)

$10,400

ouch -- too rich for my blood, even with dozens of people contributing.

[GarryN]

The information would be immensely valuable and amusing and whoever did it would get a ton of publicity. Whether it would have any effect on the way coins are graded, I dont know.

But even though only 15 coins were tested and 11 written about in the Nov issue of CW, no copper, the results are quite remarkable, when you have coins that are graded AU by a service and MS64 or 65 by another. Someone is asleep at the switch.

As a related point, I have to have my 3 Leg Buffalo photographed by Chuck, by the way, where is Chuck? It is graded AU55, but the nicest looking 3 Legged Buffalo I have ever seen. I have been offered $1,000 for it.

[Mark]

The analysis of the grading services was, in my opinion, useless. And it was useless even beyond its small sample because ANOVA won't work for what the writer was trying to accomplish.

Take this example: Suppose there are 9 grading companies that grade every coin, regardless of "actual" grade, VG8. Suppose you have a bunch of coins ranging from AU50 to MS67. Submit these coins to them and PCGS (or NGC or ANACs or whomever.) The 9 companies will return each coin graded VG8; PCGS will return the coins with a number of different grades. The Coin World author's analysis will be for this sample that PCGS's grades are the most liberal (because 9 of the 10 graded each coin as VG8) and also the least consistent (because PCGS's grades varied whereas the other grades were all consistent). But obviously that analysis is nonsense because the grades by the 9 companies are nonsense.

This extreme case makes clear that ANOVA won't work in this sort of situation and that's true regardless of the sample size. The right way to conduct a test is to submit the same set of coins repeatedly to different companies to get a measure of each company's consistency and also a measure of how the average grade for each particular coin compares across companies. But, as was pointed out, that would be expensive and I, personally, would rather put my dollars into coins rather than testing coin grading companies.

Mark

[GarryN]

Mark, could you please explain the term "ANOVA" ? I did not see it used in the Coin World article. Thank you.

Such a test of grading firms, as Mark has described would only highlight what we already know about grading. It is an art and not a science. Dave Bowers, in his column in the same Nov 17 Coin World, makes a point that he has made repeatedly.

"The printed grade on a slab serves as a great starting point from which to study other spects that are equally or more important, including quality of strike, eye appeal and aesthetic value."

See the Lincoln Cent section, 1922 Plain cent thread, for pics of the subject coin that was crossed from F12 to VF20. Enough to give a guy a heart attack.

[Mark]

GarryN:

Sorry about that--ANOVA stands for "Analysis of Variance" which is the method used in the Coin World article.

I also concur wholeheartedly with your reference to Bowers' comment that the pritned grade is a good starting point. I think you'll agree that for some third party grading companies the starting point is better than for others!

Mark

[GarryN]

Yes, Mark, I agree... I collect Morgan Dollars and I think one of the grading criteria is the phases of the moon. It doesnt matter which one grades it.

[GarryN]

Mark: Your opinion about the analysis of the grading companies was vindicated in a scathing editorial in the Dec 8 Coin World. The editorial was written by a statistician.

[smed]

What a wonderful guest editorial.

But... that means PCGS isn't the most conservative and ACG isn't the most accurate as was determined by the statistics...

...like anybody in their right mind would think those thoughts...

[GarryN]

Yes, isnt honesty a refreshing thing!

[Mark]

Garry:

I read the editorial and generally agree. There were two points about the editorial I disliked: First, a lot of the emphasis was on "garbarge in, garbarge out" whereas I think there were other major problems. Second, if I had written an editorial I would not have been as blunt about the uselessness of the initial analysis. I see no reason to be that blunt because I presume that the author of the initial analysis has human feelings! I don't mind being blunt on these boards because I doubt the author will encounter my comments, but in Coin World it's 100% sure that the initial author will see the critical comments.

Mark