I have discovered an error in the new MWDIST(x, n, tails) function. This error only occurs rarely; in particular when C(n+1, 2) is even and x = C(n+1,2)/2.
In such cases, MWDIST(x, n, 2) > 1.0 and MWDIST(x, n, 1) > 0.5. The fix is simple: when tails = 2, then simply subtract the value output by the function from 2.0 and when tails = 1 then subtract the value generated from 1.0.
E.g. =MWDIST(7, 14, 2) produces the value 1.0625 (here C(7+1, 2) = 28 = 14*2), whereas the correct answer is 2.0 – 1.0625 = .9375. Similarly, =MWDIST(4, 5, 1) produces the value 0.5625 (here C(4+1, 2) = 10 = 5*2) , whereas the correct answer is 1.0 – 0.5625 = .4375.
This error will be corrected in the next release of the Real Statistics software.