I think I may have found a possible bug in the 'DTK' package. Are you the correct person to contact about this?
In regards to the calculation of the CI term, consider the example
A = {2.857,3.098,3.077,3.012,2.978,3.234}
Vstar = {39.856,206.674,140.216,59.298}
Your code gives for CI
CI = {44.868,38.346,28.455,57.698,50.522,43.463}
However, calculation by hand based on the original paper yields
for (vstar_i + vstar_j)^.5 = {15.701,13.419,9.958,18.625,16.309,14.125},
and
CI[1] = 15.701 * 2.858 = 44.868
CI[2] = 13.419 * 3.098 = 41.571
CI[3] = 9.958 * 3.077 = 30.640
CI[4] = 18.625 * 3.012 = 56.091
CI[5] = 16.309 * 2.978 = 48.560
CI[6] = 14.125 * 3.234 = 45.686
As another indication consider the vector A has length (# groups choose 2),
but during the calculation of CI only the first (# group) levels of A are looped through.
I apologize if I missed something or this is an old version but I thought I would bring this to you attention.
Sent using CloudMagic