View Single Post
Old 06-15-2010, 06:43 AM   #418
beppe
Grand Sorcerer
beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.beppe ought to be getting tired of karma fortunes by now.
 
Posts: 5,161
Karma: 81026524
Join Date: Feb 2010
Location: Italy
Device: Kindle3, Ipod4, IPad2
Quote:
Originally Posted by kennyc View Post
There is only ONE scientific method. Again you seem to be confusing practice with theory. The Scientific method goes thus:

1. Create a falsifiable hypothesis based on experience or existing knowledge
2. Design a repeatable, replicable experiment to test the hypothesis
3. If the hypothesis passes the test (and is repeatable and replicable) it is deemed to be true and is now a theory describing the particular behavior or activity etc.
If the hypothesis fails the test then it's back to step 1.
Dear friend Kenny,
lets put some pepper into the discussion. (Vinegar?) Of the several objections that come to my mind I will choose step 3 in your very clear statement.

To perform the test one has to measure the discrepancies between the observations of the experiment and the consequences of the hypothesis.
(I use these 3 names so that I can address them simply. Their meaning should be obvious and undisputed, being contained in your statement).

A) To obtain the discrepancies you have to build some generalized comparing procedure, algorithm, call it whatever you want but you need it, and you need to define it very clearly. You see, comparing 3 with 1.7 is quite easy: the algebraic discrepancy is just the difference 3 - 1.7 = 1.3 But then somebody jumps up and say. Nah, I do not like algebraic discrepancies, positive and negative compensates. Use least squares. And other one objects, nahhh it is too smooth and it is fragile versus outliers. Use something more robust. and on and on. Of course the actual discrepancy you get it depends on how you build it. There goes objectivity. There enters black magic.
B) then you have to perform a measure of the discrepancies. That is reduce an entity that can be quite complicate to a number, a number that will be used in the test. Oh, here things start to get very complicated, and again quite open to any bending for devious purposes.
C) the test. Ah,Ah. Every head is own test. Parametric, non parametric,...

So maybe your spiel should not be taken literally, nor in a strict sense?
Aha!

This was nasty enough.

I have worst things to say about other points.

I am going to have my lunch.
beppe is offline   Reply With Quote