statsmodels.stats.oneway._power_equivalence_oneway_emp

statsmodels.stats.oneway._power_equivalence_oneway_emp(f_stat, n_groups, nobs, eps, df, alpha=0.05)[source]

Empirical power of oneway equivalence test

This only returns post-hoc, empirical power.

Warning: eps is currently effect size margin as defined as in Wellek, and not the signal to noise ratio (Cohen’s f family).

Parameters:
f_stat : float

F-statistic from oneway anova, used to compute empirical effect size

n_groups : int

Number of groups in oneway comparison.

nobs : ndarray

Array of number of observations in groups.

eps : float

Equivalence margin in terms of effect size given by Wellek’s psi.

df : tuple

Degrees of freedom for F distribution.

alpha : float in (0, 1)

Significance level for the hypothesis test.

Returns:

pow – Ex-post, post-hoc or empirical power at f-statistic of the equivalence test.

Return type:

float