Testimonial and Anecdote as Evidence

A testimonial is an uncontrolled personal account of what happened to someone after they took something. It is not a weak version of a trial; it is a different kind of object. A trial is built to answer a counterfactual question, what would have happened to the same people without the intervention, and does so with a control group, randomisation, blinding and prespecified outcomes. An anecdote has no counterfactual at all, so the observed change is the sum of the drug effect and everything else that was happening at the same time, with no way to separate the terms.

Those other terms are large and well characterised. Placebo responses in subjective outcomes such as pain, fatigue, mood and libido routinely reach a substantial fraction of the response seen on active treatment. Conditions that fluctuate produce apparent improvement whenever measurement starts at a bad point, which is precisely when people start a new compound. People who begin an intervention usually change other things simultaneously, and recall of a baseline is reconstructed after the fact by someone who now has an opinion.

The population of visible testimonials is also filtered. Those who felt nothing rarely post, those who were harmed often stop participating in the community, and where sellers curate or commission the reports the sampling frame is chosen to be favourable. There is no denominator, so a hundred enthusiastic reports are consistent with a hundred responders out of a hundred or out of ten thousand.

Anecdote does carry real evidential weight for one thing: an unexpected, severe, biologically striking harm. Case reports of limb malformations in 1961 identified the thalidomide signal long before any controlled study. That asymmetry is the rule worth remembering, since anecdote detects dramatic harm well and modest benefit not at all.