Before starting an A/B test, it is recommended to first do an A/A test to make sure that we get a 50:50 ratio of observations into 2 groups and the 2 groups are not different. W