As many as was required to find statistical significance. This S in BLS stands for statistics, after all.
What are the flaws in this methodology?
What are the flaws in this methodology?