statsmodels.stats.weightstats.ttost_ind

statsmodels.stats.weightstats.ttost_ind(x1, x2, low, upp, usevar='pooled', weights=(None, None), transform=None) [source]

test of (non-)equivalence for two independent samples

TOST: two one-sided t tests

null hypothesis: m1 - m2 < low or m1 - m2 > upp alternative hypothesis: low < m1 - m2 < upp

where m1, m2 are the means, expected values of the two samples.

If the pvalue is smaller than a threshold, say 0.05, then we reject the hypothesis that the difference between the two samples is larger than the the thresholds given by low and upp.

Parameters:

x2 (x1,) – two independent samples, see notes for 2-D case upp (low,) – equivalence interval low < m1 - m2 < upp usevar (string, 'pooled' or 'unequal') – If pooled, then the standard deviation of the samples is assumed to be the same. If unequal, then Welsh ttest with Satterthwait degrees of freedom is used weights (tuple of None or ndarrays) – Case weights for the two samples. For details on weights see DescrStatsW transform (None or function) – If None (default), then the data is not transformed. Given a function, sample data and thresholds are transformed. If transform is log, then the equivalence interval is in ratio: low < m1 / m2 < upp

Returns:

pvalue (float) – pvalue of the non-equivalence test t1, pv1 (tuple of floats) – test statistic and pvalue for lower threshold test t2, pv2 (tuple of floats) – test statistic and pvalue for upper threshold test

Notes

The test rejects if the 2*alpha confidence interval for the difference is contained in the (low, upp) interval.

This test works also for multi-endpoint comparisons: If d1 and d2 have the same number of columns, then each column of the data in d1 is compared with the corresponding column in d2. This is the same as comparing each of the corresponding columns separately. Currently no multi-comparison correction is used. The raw p-values reported here can be correction with the functions in multitest.