I'm talking here about statistics as a mathematical discipline.
When you look at something with experimental means (How tall is a person? What did God tell me in that dream?) you never get all the information. You get a statistical sample.
If you measure the height of 100 random people, does that tell you how tall on average the people of a whole country are?
If you add up what 100 random prophets tell you about God, does that give you the ultimate and infallible truth what God is like?
No, it doesn't.
The larger the sub-set, the closer the information you get from it is to the information you would get from the whole set.
But you will hardly be able to measure the height of ALL people in a country.
You will hardly be able to collect the wisdom of ALL people who would like to talk with you about God. (And try combining them into one coherent explanation/theory...)
The scientific method is based on this simple mathematical fact. No matter how much proof you gather in support of your theory, there is always a probability that something somewhere in the universe violates the theory you made up. (Unless you took the time to gather all possible data in the whole universe.) And this one instance is entirely enough to make your theory deviate from the "true" theory that you are trying to decipher.
Religion works exactly the other way round. No matter how little proof you have in support of your theory, it is defined as impossible that your theory could be wrong. You will eventually find some or other evidence that agrees with your theory. So, you have that speck of data supporting your religious theory and shit-tons of data contradicting your theory. In the realm of religion this means that your theory is correct.
EDIT:
THIS HAS NOTHING TO DO WHATSOEVER WITH THE PERSON COLLECTING OR ANALYZING THE DATA. THE POSSIBILITY OF ERROR IS AN INTRINSIC MATHEMATICAL PROPERTY OF THE CONCEPT "DATA".
= new reply since forum marked as read