Hi, Chris. The idea here is that you check for suitable sample size and keep generate it until it reaches a good significance level. However, everything will be more clear in the article I’m writing, which will explain how to build a good sample without trial-and-error.