Talk:Siegel–Tukey test

power

i probably an error, at the end of the article, : "power is equal to 0.61%." it is probably more some thing like 61%, no ? al

math for U1 and U2 wrong

I don't know what the correct formulas are but the simple math for U1 and U2 seem to wrong as the calculation of the probability (if 9 is correct correct statistic) This math needs correction or deletion.

cause of this error?

I guess the cause of this error.

At first, when the number of all observations (N) is odd, the middle observation is thrown out. So, the seventh element, 66 in group B, is thrown out, and the number of observations in the group B is five (n=7, m=5, and N=12).

Then, the sum of the ranks within each group and the U statistics are computed as follows:

W_a = 5+12+11+10+7+6+3 = 54
W_b = 1+4+8+9+2 = 24
U₁ = 7*5+(7*(7+1)/2)-54=9
U₂ = 7*5+(5*(5+1)/2)-24=26

Thus,

U₁*U₂ = 9+26 = 35
n*m = 7*5 = 35

Additionally, I confirmed that U₁+U₂ and n*m are equivalent when N (or n+m) is even, because U₁+U₂=N(N+1)/2.

Is my guess is correct?

[1] J. D. Gibbons and S. Chakraborti, "Nonparametric Statistical Inference", Marcel Dekker Inc., p.330, 2003. (We can peek this page on Google :))

160.252.131.12 (talk) 11:10, 22 April 2009 (UTC) (only the section "cause of this error?")

As I like to say, our friend's English is better than my Italian. Nevertheless, I have rewritten the article to make the English more comprehensible. I have also clarified the U statistics. I did have to remove a couple of sentences that I could not understand at all. Markjoseph125 (talk) 17:58, 2 August 2009 (UTC)