| Author |
 Replies: 40 / Views: 4,381 |
Page 3 of 3
|
|
|
|
Pillar of the Community
United States
850 Posts |
Here are my results
Nickels: $672: Silver: .5 Oz (10 wartime) .75 oz per $1000
Dimes: $760: Silver: .63 oz (9 silver) .82 oz per $1000
Quarters: $670:Silver: .18 oz (1 silver) .268 oz per $1000
Half:$1500: Silver: 15.6 oz (94 40% and 4 90%) 10.4 oz per $1000
Halves are by far more productive since I started keeping track this May.
|
|
Pillar of the Community
United States
842 Posts |
|
|
Rest in Peace
United States
9104 Posts |
Quote:
I have to disagree with those who have said you will probably find the most silver in halves in this experiment. I would have agreed with this 5 years ago or so, but I find that halves are so searched through anymore. A quick post count of various threads (half finds, dime finds, etc) confirms this. |
|
Valued Member
United States
58 Posts |
I can't wait to see the results. Sometimes I'll get out 10 boxes and not find a single one. Then I take out a couple and find loads. That makes it fun - it's a surprise!
|
|
Bedrock of the Community
Australia
21788 Posts |
Halves have the least amount of circulation.
That may be true.
As well, it also may mean that halves are searched less than they are circulated.
|
|
Pillar of the Community
United States
7629 Posts |
Saruma - I was only trying to figure out whether any issues (especially in early zinc cents) were prone to attrition more so than earlier or later counterparts, and to attempt to be able to assume the ratio of large date to small date cents in the dates where they existed.
bibd - the actual 'useable' data was such a small sample, none of it mattered in the end. Enough people couldn't follow directions to make the entire project useless.
|
|
Pillar of the Community
United States
548 Posts |
Quote:
I tried this with Lincoln cents a few years back - trying to see what was out there in circulation. I found that even though I had results on tens of thousands of coins, I wasn't covering a statistically significant percentage of circulation overall to make the statement that I knew what was in circulation. For Lincoln cents the number of coins I would have to cover would be over 300,000,000 to ascertain that I had (or my volunteers had) seen enough of all the coins in circulation to make even a good guess as to how many of one issue were out there versus another. This would have required that 60,000 people volunteer to look through a full 5,000 count bag of coins and sort them by date and mintmark and report the results. Then it would have required that I somehow store all this data so it would be retrievable - on no budget. It was way too much of a project to handle and get any useful numbers.
I know this project is different, but one comparison I can make is that a thousand bucks from one location is not going to provide even a hundredth of the coins necessary to make any useful statistical analysis. You'll have to be into the millions of dollars worth searched before your numbers will even be close enough with which a wild guess could be based. Gosh, I may be the only one to disagree. Has anyone done the statistical analysis to see how many coins must be sampled to calculate this? Statistics is a funny thing; sometimes small sample sizes yield "perfect" results. BTW, I haven't done the analysis. I hated statistics. |
|
Pillar of the Community
United States
1903 Posts |
Based on a population of 460 billion coins in circulation and 2500 coins in a box, you would need roughly 385 randomly acquired boxes of cents from circulation to get a 95% confidence level with an accepted error of +/- 5%. If you wanted to be really precise and have a 99% confidence with +/- 1% error you would need to randomly aquire a little over 16,600 boxes. So if you were willing to accept that if you acquired the 385 boxes you would have 90% chance of getting a representative sample of what the current circulation population looks like. This is a doable task.
In reality there aren't still 460 billion cents still circulating. This number was achieved by simply adding up total mintages. That being said this statistical sample size is actually larger than it would need to be, but only gives a better confidence in the sample size stated.
Anyone up for trying to acquire 385 boxes of cents randomly across the country on a single day? :-)
|
|
Pillar of the Community
United States
548 Posts |
|
|
Bedrock of the Community
Australia
21788 Posts |
From my visit to the 'States, I had suspected that the Cent may be a single use coin only. Not quite true perhaps, but certainly less so than the other denominations.
It seems to me when states taxes are applied at the point of sale, the final combined sale price is expressed exactly in odd cents amounts, and that is what is paid. Folks just take these leftover cents home and just hoard them, because they are just too much trouble to bother with. When you have got lots of them, you just take them back to the bank for recycling. That's just more work for the bank with no benefit to anyone.
This phenomenon is much more common in the 'States than anywhere else that I have been. The result is that more Cents are needed, and they have much less circulation per coin. There is also a greater proportion (numerically) of newly minted Cents to be found in circulation.
A statistical analysis should prove this postulation.
Edited by sel_69l
10/22/2011 9:06 pm
|
|
Pillar of the Community
United States
564 Posts |
Also your random sample would need to be from all over the country. Your findings are only for your area. Your sample size must be larger to get an accurate percentage.
|
|
Rest in Peace
United States
9104 Posts |
Quote:
Anyone up for trying to acquire 385 boxes of cents randomly across the country on a single day? Not that easy. Boxes tend to be circulation "returns" filled in with new coins as needed. You'd never find a box of 2011 coins in a random grouping of cents. You might if you buy boxes. You'd need 385 box equivalents from Coinstars, Unicef and door-to-door collections by boy scouts. |
|
Pillar of the Community
United States
627 Posts |
Ok, I have to chime in here. Firstly to give a "disclosure", I am a practicing professional statistician with almost 10 years experience, and with expertise in sampling and study design methodology. I have a bachelors degree in mathematics and a masters degree in statistics. Please don't hold these things against me, as I will give my 2 cents regarding your "design" and plans. 1. Your sample does not need to be a certain percentage of the whole of all coins still in circulation or ever minted. Someone suggested that they needed to sample 300 million coins, which is absolutely absurd. Think of it as you would a random telephone survey regarding opinions. There are ~308 million persons in the U.S. today, yet all phone surveys are usually based on a "random" sample of 1000 persons. Some 60 years ago when there were about half as many (150 million) people in the US, all surveys were still conducted on 1000 people. Why 1000? Because a sample of 1000 persons give you estimates of opinions with a margin of error of roughly +/- 3%. The size of the total population from which you sample does NOT matter; what matters is that your sample be as "random" as is possible to represent the universe to which you are trying to make references. 2. The main problem I see with your plan is the source of your coins. As justcarl and Conder pointed out, the mix of silver coins from different geographic areas will likely differ. And, as someone else pointed out, if you get your coins from a source with little "churn" you will not be able to make certain you are not searching the same lot of coins unless you keep them all in your possession until you have completed the study. The best design would be to get friends or persons from MANY different geographic locations across the U.S. to each search a fixed number of coins. Given that you expect the percentage of success to be very small, this will push up your needed sample size to get a stable estimate. 3. Another problem of your "study" is that you are trying to estimate a moving target. I suspect when you find silver you are removing it from circulation/bags. In addition, there is a non-random influx of silver coins that make it into those bags. You cannot know how many other people have already picked over your bag of coins. Thus, even if you get a geographically diverse sample of estimates as suggested in point (2) above, you will not be able to quantify the "variables" like a) number of collectors in the area who search coins, b) amount of "churn" of coins at that particular bank/branch (are their coins coming in from people's coffee cans or from the local armored car service, etc.), and many others (too numerous to list). SUMMARY: I would not suggest undertaking such a "study". There are far too many variables you can not control, and thus your results will have a scope of inference to essentially no population. The use of "statistics", per se, is not your problem. The primary and unaddressable problem is the design of the study. As we say in research -- Garbage in, garbage out. You will have spent a huge amount of time coming up with an answer to a non-existent question.  |
|
Pillar of the Community
United States
627 Posts |
If you don't want to read my previous long response, I will just say this. I am a professional statistician, and what you have is not a statistical problem, but a sampling bias problem. You can calculate all of the statistics you want on your data, but the fact that you have a non-random sample will make all of your inferences meaningless to the universe of coins at large. |
|
Pillar of the Community
United States
548 Posts |
Cool.  |
|
Page 3 of 3
|
Replies: 40 / Views: 4,381 |
Page 3 of 3
|
First Page
Previous Page
Posted 10/17/2011 4:45 pm
