Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.


Luis A. Afonso
Posts:
4,758
From:
LIsbon (Portugal)
Registered:
2/16/05


Attention: an error concerning the KolmogorovSmirnov Test
Posted:
Sep 1, 2012 4:07 AM


Attention: an error concerning the KolmogorovSmirnov Test
A Modified KolmogorovSmirnov Test for Normality [Drezner . . .] mpra.ub.unimuenchen.de/14385/
This paper shows an error, pg. 3, formula (1). We are dealing with the maximum difference between the ABSOLUTE VALUE of k/n  Fi((x_k  mu)/sigma) and the ABSOLUTE VALUE of Fi (x_k  mu)/sigma)  (k 1)/n. If the absolute values are not taken into account we could be before a very large negative difference, say 1 compared with a small positive, +0.0001. If we use the formula as it is displayed in the text  raw values, not absolute ones  we will choose the latter which is an absolute absurdity because the job concerns DISCREPANCIES: the larger one among all experimental points. I really wish that it was merely a formula misprint; the calculations were really performed as expected.
Luis A. Afonso



