Lies, damn lies and Statistics ...

My statistics guru used to say, there are three types of lies. Lies, damn lies and statistics. But since Statistics is the chosen method to differentiate between a signal and a noise (declaration of any event), we will have to bring it in to evaluate most things.

I have received a lot of emails on my post about the Astrology using your thumbprint. I had suggested that maybe we need an extensive database and a large sample size to see if there is some hidden formula for lifes events compared to thumbprint features.

On an interesting note, my buddy (yes, my chai partner) who happens to be a Ph.D in particle physics told me some things. They sit in trailers along the tangent of a giant underground tunnel (called a particle accelerator) where sub atomic particles are made to circle the tunnel (the diameter of this tunnel is a few miles, I am not kidding.. I have been to Fermi Lab three times to visit him) and then they are crashed into other particles and the by products are sucked into these tangents on the tunnel using special electric and magnetic fields. I am sorry to bore the average reader with this depth of scientific detail. Lets get to the point .

Here is a conversation we had a long time ago:

me : how many more years are you going to keep doing this thing. I have already finished my degree and I am going to work. Your parents are getting impatient right ?
dost : it is not that straighforward macchhi. i am trying to prove that a certain particle reaction occurs. (when particle A hits B we get particle C + something !!)
me : why does it take that long then ? just find it.
dost : ha ha.. the tunnel is running 24/7 and the data is also being collected 24/7. But most of the time the common well known reactins predominate. The rare ones happen one in 10000 times !
me : You are saying 24/7. I can also do some math. So you should have had this reaction at least once in the last 5 freaking years you have been sitting in this trailer right ?
dost: yes. but just by showing that reaction happened once, no one will give me a degree. I have to show that it happens consistantly (?!). I have to show that it happens at least a 10000 times. That means after collecting 10 billion events, I have to filter out this event and show it happens 10000 / 10 billion times to prove it !!
me : good luck. I will come back a few decades from now and hopefully you will have graduated.

(the guy did graduate and is on his way to a professorship !)

But there you have it folks, sample size matters. If you look at one feature on a thumbprint and associate it with say a short life expectancy, we have to compare that same feature and its variations over a large enough sample size and then maybe we can determine the confidence level in that prediction. I also recieved a lot of mail on how astrology is actually all mathematics. I can see how astronomy is mathematics, and how some parts of astrology like horoscope determination is mathematical, but the predictive part being math is hard for me to believe. That does not mean it is not mathematical, just that my small brain goes into overload at that thought. If you know of any treatise, thesis on the mathematical nature of predictive astrology, please forward it to me !!