scipy.stats from the ground up: is the difference between two samples real?

Afterwards you can describe two samples with scipy.stats, test whether they differ, and report the effect size with a confidence interval, not a bare p-value.