FWIW, I think it may be statistically-significant eyeballing the graph. Activities, being counts bounded at 0, can't be normal, and these are very non-normal data, so the t-test is wrong; if he wanted a p-value, https://docs.scipy.org/doc/scipy/reference/generated/scipy.s... would've been more appropriate, and I'd bet the p-value will be noticeably smaller.