3.11.18.2.1. statsmodels.stats.proportion.AllPairsResults¶
-
class
statsmodels.stats.proportion.AllPairsResults(pvals_raw, all_pairs, multitest_method='hs', levels=None, n_levels=None)[source]¶ Results class for pairwise comparisons, based on p-values
Parameters: pvals_raw : array_like, 1-D
p-values from a pairwise comparison test
all_pairs : list of tuples
list of indices, one pair for each comparison
multitest_method : string
method that is used by default for p-value correction. This is used as default by the methods like if the multiple-testing method is not specified as argument.
levels : None or list of strings
optional names of the levels or groups
n_levels : None or int
If None, then the number of levels or groups is inferred from the other arguments. It can be explicitly specified, if the inferred number is incorrect.
Notes
This class can also be used for other pairwise comparisons, for example comparing several treatments to a control (as in Dunnet’s test).
3.11.18.2.1.1. Methods¶
__init__(pvals_raw, all_pairs[, ...]) |
|
pval_corrected([method]) |
p-values corrected for multiple testing problem |
pval_table() |
create a (n_levels, n_levels) array with corrected p_values |
summary() |
returns text summarizing the results |