Numerical simulation of random samples of fish

[hdtran]

I have a grant report which I'm procrastinating on... So, I wrote a short MATLAB program which simulates the following:

1,000,000 damsels, of which 10% are tainted
250,000 tangs, of which 40% are tainted
150,000 angels, of which 20% are tainted

The program then mixes all the fish up (all 1,400,000 of them), and randomly samples about 384 fish (not exactly 384, because I'm too lazy to write the code to make it do that). It then calculates the % which were tainted (doesn't care if it's a tang, damsel, or angel that it tested), and also calculates the actual percentage tainted of the 1,400,000 total fish.

Each time you run this, you will get slightly different numbers, because it uses a random number generator (like tossing a dice) in the computer.

A typical result is:
380 fish sampled; 65 tainted; 17.1% bad
Actual bad percentage was 16.4%

Kalk, I will offer to do the following for you (within reason):
Tell me:
(1) How many damsels, and what % are bad
(2) How many tangs, and what % are bad
(3) How many angels, and what % are bad

I will run the simulation once, and report the results (just total fish which tested bad and total fish tested, versus the actual number).

My computer greatly appreciates it if you can keep the total number of fish in the one million (may be able to do a few million) range. (I don't have Gigs of RAM...)

Regards,

 

[Kalkbreath]

Its not that simple .....see with your sampling you have all the fish to choose from..............Peter did not have all the 25 million fish in a swimming pool and then sample from all the 25 million fish . What ever conditions that led to the collection of fish candidates for testing .....effected the results . Its like selecting from a deck of cards ....with some of the cards missing .

 

[hdtran]

Fair enough.

I'll make 25 million fish, and mix them all up. Then, I'll throw away 20 million, and pull about 384 fish from the remaining 5 million.

Want to try that?

Hy

 

[Kalkbreath]

No, because when you "make " the fish" you will literally be making up fish. What percentage of handguns are used to defend a person each year? Because ninety-nine percent of the time when a hand gun is used for defense the firearm is not discharged. You are never going to be able to determine the answer {how often a gun is pointed but not fired } by studying the data of firearms that have been discharged. You have no numbers or data to "plug in to your statistical formula. The findings are only as accurate as the input data. You have confused my complaint that Peters study contained too few specimens .......from the greater point of too few collectors and the data not being chosen from the entire fish "pool". Lastly.....if you want to use you formula ....then ask Peter what the second greatest number of any one fish type {species} was included in the data for any given year. Then take that number and use it against the total fish exported of that species for the year. Example, Green chromis.....forty sampled .....Twenty-thousand exported for the year 1998. Is there any species of fish that was included in this study which posted enu\ough specimens for statistical computation?