Ask an analyst how precise their method is and they will usually quote a number from their own hands: five injections, one afternoon, a relative standard deviation (RSD, the scatter as a percentage of the average) under 1%. It is a true number. It is also the most flattering one the lab will ever produce, because it leaves out everything that changes when the work is shared: the second analyst, the other balance, the Tuesday column, the standard made last month by someone on nights. In most labs a calibration curve is not one person's work. It is built, checked, reused and relied on by a team, and the precision that matters is the team's.
Three kinds of spread
Run the same sample several times and the results scatter. How much depends on what changes between runs.
If the same person runs it with the same equipment on the same afternoon, the tight spread you see is called repeatability. If a different laboratory runs it, the much wider spread is reproducibility. Between the two sits the one a working lab lives with. When the same lab runs the method on different days, with different analysts and different equipment, the spread is wider than repeatability; that spread is called intermediate precision.1 It is at least as large as repeatability, often larger, and it is the precision that matches how the lab actually reports: a result from whoever was on shift.
The same pattern shows up one level higher: collaborative studies, which send the same samples to many laboratories, exist precisely to measure the extra spread that appears between labs.2 The general lesson carries down into a single lab: every source of variation you leave out of a precision study makes the number smaller than the truth.
So the precision study that matters is one where the people who run the method each prepare and run the same standards across several days. Validation guidance such as ICH Q2 describes how intermediate precision is studied when a method is validated, and your quality system says how it is watched afterwards; the point here is simply that "analyst" is a real source of variation and belongs in it. Small differences in technique (how a syringe is rinsed, how a bubble is cleared, how long a flask is left to reach temperature) are real effects, and they only show up when more than one person does the work.
This lesson belongs to From signal to result, a course on the gas chromatograph from detector to reported number, coming soon. The rest of the lesson, its questions and the course run will be part of it.
Want to hear when it opens? Follow The Analytical Column below, or start with the free lessons: Start here →