They should use Bayesian statistics, in which case it doesn’t matter when you stop (more precisely, stopping when you get a result does not bias the outcome; of

They should use Bayesian statistics, in which case it doesn’t matter when you stop (more precisely, stopping when you get a result does not bias the outcome; of course running the test longer will make the result more robust). See http://andrewgelman.com/2014/02/13/stopping-rules-bayesian-analysis/ http://andrewgelman.com/2014/02/13/stopping-rules-bayesian-a...