When running an A/B test in python with scipy stats, we haven argument call: equal_var.

More info here: https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.ttest_ind.html

But basically it means: If True (default), perform a standard independent 2 sample test that assumes equal population variances [1]. If False, perform Welch’s t-test, which does not assume equal population variance [2].

I wonder, if A & B are from the same population before the start of the experiment, we should set this parameter as True, right?

I have seen that a lot of examples set them as False, but it seems that is more when comparing means of two distribution, not necessary from the same population, in which case, makes sense.

Is this correcto?

  • $\begingroup$ Thank you! A little... The thing is that I would like to understand on the context of an A/B testing. Where before the test, both samples came from the Equal variance. @JeremyMiles $\endgroup$
    – marz
    Jun 14 at 13:08
  • $\begingroup$ I'm not sure what you're asking. Equal variances is an assumption that usually doesn't matter very much, and when it does is easy to correct. Also set it to False. There's no reason not to. $\endgroup$ Jun 14 at 15:42