TFA says multiple times that the results are affect by random chance
Yes, but recognising that is only the first step. Quantifying the variance is the next step which I miss in the article.