# Self-Selected Testimonials Cannot Establish an Effect Size ## Claim Letters from self-selected, already-persuaded listeners can illustrate that a practice worked for someone and cannot establish how well it works in a population, regardless of how confidently they are framed. <!-- atomicity: 1 idea ✓ --> ## Scope & conditions - The selection happens twice — who writes in, and which letters get read — and both filters run toward success. - Illustration is a legitimate use; the error is the slide from "here is what it looks like" to "here is how well it works." - No number of testimonials fixes this, because the sample's bias does not shrink with its size. - Applies to case studies, customer quotes, and my own "this worked for me" notes. ## Evidence From Cal Newport's episode on reversing brain rot: three listener letters are used to support a general claim about the practice's effect. ## Links - [[Named Brain Structures Without a Citation Are Rhetoric]] — the companion evidence defect in the same episode, on the mechanism side - [[Pair One Named Case With a Population Statistic]] — the constructive fix — keep the case for salience and get the warrant elsewhere - [[Genre-Wide Claims Need More Than One Catalog as Evidence]] — the same sample-versus-population error in a different domain - [[3. Resources/MOCs/Evidence & Claims Literacy MOC]] — the claims-literacy thread on sample versus population.