----- Original Message -----
From: <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Thursday, July 05, 2001 8:50 PM
Subject: Re: Earth on Edge
> It is important to understand that the answers to these questions lie
within
> statistical studies and experts can do have honest differences on the
> interpretation of the same data set. As Steven Pinker likes to say humans
> are bad natural statisticians. Let me give you an example: Which of these
> statements is more impressive.
> 1) there was three times the expected rate of a specific cancer in a small
> town of 1000 for three consecutive years
> 2) There were three additional cancers per 100,000 lives in one year in a
> population of 3,000,000.
>
It depends on the numbers of cancers as you know. But, I know of a
marvelous cheat: Monte Carlo techniques. It allows you to state your
assumptions up front, and then turn the crank and get your answers. When I
was in grad. school I used these techniques to do error analysis on my data.
Later, a post doc tried to teach me the analytical approximation for the
same calculation. After listening politely for 10 minutes, I asked, "but
isn't this an approximation; wouldn't a Monte Carlo give a better answer?"
He said, "Yes, but do you know the cost of the computer time to run such a
Monte Carlo?"
I said, yes, 28 cents. I ran it last week.
He looked at me and said
"grad students have it too easy nowadays."
Seriously, if you are interested in using these techniques I've got a
zillion years of experience in taking the easy way out that I'd be happy
to share.
Dan M.
