What proportion of simulated statistics are at least as extreme as the observed statistic obs_mean_diff? In other words, what is the (simulation-based) p-value?

A) 0.01
B) 0.05
C) 0.10
D) 0.25