Siegel S, Castellan NH.† Nonparametric statistics for the behavioral sciences. 2nd ed New York: McGraw-Hill, Inc; 1988,
post hoc comparisons after Friedmansí test .† pages 180-183
post hoc comparisons after Kruskal-Wallis test .† pages 213-215
Tables: pages 319-321
The formulas 1a and 1b depict for post hoc comparisons after Friedmansí test .†
In the formula Ia, |Rv - Ru|† represents difference in rank sums for a particular pair, k represents the number of groups† and the critical value for Za/k(k-1) is the abscissa value from unit normal distribution above which lies a/k(k-1)which is obtained from relevant statistical tables.† Similarly, in the formula Ib, the |R1 - Ru| is the difference in rank sum between the control group and the group to be compared, c is the number of comparisons i.e k-1.† The critical values for q are obtained special table.
The formulas IIa and IIb depicted below represent those for post hoc analysis after Kruskal-Wallis test for all pair-wise comparisons and comparisons against a single control group.
In the formula IIa, the critical value must exceed the difference in average ranks between a pair of groups to be compared, N is the total sample size for all the groups, nu and nv represent the sample size of the relevant groups to be compared.† The critical value† Za/k(k-1) is the abscissa value from the unit normal distribution above which lies a/k(k-1) percent of distribution.† The values of z can be obtained from relevant statistical tables.†
†In the formula IIb, the difference in average ranks between control group and the group to be compared must exceed the critical value and nc represents the sample size of control group.