statsmodels.stats.weightstats.ttost_paired(x1, x2, low, upp, transform=None, weights=None)[source]

test of (non-)equivalence for two dependent, paired sample

TOST: two one-sided t tests

null hypothesis: md < low or md > upp alternative hypothesis: low < md < upp

where md is the mean, expected value of the difference x1 - x2

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.


first of the two independent samples


second of the two independent samples

low, uppfloat

equivalence interval low < mean of difference < upp

weightsNone or ndarray

case weights for the two samples. For details on weights see DescrStatsW

transformNone or function

If None (default), then the data is not transformed. Given a function sample data and thresholds are transformed. If transform is log the the equivalence interval is in ratio: low < x1 / x2 < upp


pvalue of the non-equivalence test

t1, pv1, df1tuple

test statistic, pvalue and degrees of freedom for lower threshold test

t2, pv2, df2tuple

test statistic, pvalue and degrees of freedom for upper threshold test