Another Reason to Abandon P-values; or Another Way to Cheat

The Sample Size Extender™ is so easy a child could use it. Here’s how it works. All you need is a theory: any will do, the wilder the better. Begin collecting your data, put them through the works one at a time, then just wait for the publishable p-value, which is sure to come.

Here’s an example. Your theory supposes that a certain measure will be different between two groups. You’ll confirm this using a t-test—though any test will do. You begin by measuring two people, one from each group1. Put the measurements of this pair into the t-test and check whether your p-value is less than the magic number. If it is: stop! Your theory is proven. Go forth and write your paper.

But if the p-value is not small enough, measure two more people, one more from each group. Then run the augmented sample through the t-test and check for a publishable p-value again.

Iterate the Sample Size Extender™ and you will always—absolutely always—find a p-value less than the magic number. Yes, sir, friends: this method is foolproof. Fools prove it every day!