This Page

stac.nonparametric_tests.hochberg_test

stac.nonparametric_tests.hochberg_test(ranks, control=None)[source]

Performs a Hochberg post-hoc test using the pivot quantities obtained by a ranking test. Tests the hypothesis that the ranking of the control method is different to each of the other methods.

Parameters:

pivots : dictionary_like

A dictionary with format ‘groupname’:’pivotal quantity’

control : string optional

The name of the control method, default the group with minimum ranking

Returns:

Comparions : array-like

Strings identifier of each comparison with format ‘group_i vs group_j’

Z-values : array-like

The computed Z-value statistic for each comparison.

p-values : array-like

The associated p-value from the Z-distribution wich depends on the index of the comparison

Adjusted p-values : array-like

The associated adjusted p-values wich can be compared with a significance level

References

  1. Hochberg, A sharper Bonferroni procedure for multiple tests of significance, Biometrika 75 (1988) 800–803.