Correcting the H test statistics for Ties In using the Krusk

Correcting the H test statistics for Ties. In using the Kruskal-Willis test, there is a correction factor that shoud be applied whenever there is are many tie: Divide H by

Solution

The corrected value of H-statistic is obtained by correcing H by ties correction factor (C_H)

which is given by:

C_H = 1 - [ summation of { T³ - T } ] / ( N³ - N )

where T = number of values from a set of ties

N = Number of values from all combined sample

Hope this helps.