proficiency testing
/ proh-FISH-en-see TES-ting /
Imagine a cooking school that mails every graduate the same mystery ingredients and asks them to send back a finished dish, then quietly judges them all side by side. The ones whose dishes land near the group's standard pass; the odd ones out get a quiet word. Proficiency testing does this for labs: an outside body sends identical samples to many labs and compares their answers to see who is on target.
Proficiency testing is an external check of a lab's competence in which an organiser distributes the same well-characterised test material to participating labs, collects their results, and scores each one against an assigned reference value, usually with a z-score showing how many standard deviations a lab sits from the consensus. It is a recurring, blind exam graded by an independent party.
It matters because a lab's own internal quality control can look spotless even while it is consistently, invisibly wrong — only an external comparison can expose a shared blind spot. The caveat is that a single good z-score is a snapshot, not a guarantee: it shows the lab handled one sample well on one day, so proficiency testing must be regular and is one piece of a wider quality system, not proof by itself.
Two hundred water labs each receive the same bottle of test water and report its nitrate level. The organiser assigns a reference value of 8.0 mg/L; a lab reporting 8.1 earns a z-score near zero (a pass), while one reporting 11.0 earns a z-score above three and must investigate.
An independent body scoring many labs on the same blind sample.
A z-score within ±2 is usually treated as satisfactory, between 2 and 3 as questionable, and beyond 3 as unsatisfactory, requiring the lab to find and fix the cause before it affects real reported results.