Date: Nov 23, 2012 6:56 PM
Author: Luis A. Afonso
Subject: A relief to the anti-NHST people. . .

This post will be a relief to the anti-NHST people

It´s easy to get a i.i.d. sample pair from the same Population and notwithstanding their mean values leading to a indisputable rejection they are equal by T test. It´s enough to from a large amount (8000) of N(0,1) data to chose the 51 smaller and the 51 larger values, then to evaluate T concerning the difference on Population mean values.
I performed 4 times and obtained (smaller-larger) for the 95% , 2 tails, left bound:
______T= -77.863, -86.345, -92.664, -107.836.
Knowing that (for 100 df) the critical values are lesser than -2.000, +2.000 which are the bounds relative to 60 df, we must conclude that the Population means are indisputably different one another, WHICH IS REALLY WRONG BY CONSTRUCTION!
Appreciate how much I am anti-NHST friend!
Indeed . . .

Luis A. Afonso