About studies and their design

A study holds everything for one piece of verification work: the design you plan before you collect, the measurements, the acceptance limits you judge them against, and every calculation and report that follows.

Each study answers one question, such as how much repeated results vary within a run, or whether a new reagent lot gives the same patient results as the current one. The question decides the design. Measuring one control 20 times in one run answers a repeatability question. Measuring it over 20 days answers a multi-day precision question. A study's design is fixed once it has data or a calculation. A different question needs a new study, even when it reuses the same file. Studies sit inside a project and are grouped by analyte so you can find them. Studies with the same analyte are never pooled.

A study moves through four stages: Set up, Data, Results and Report. Set up comes first because the limits are part of the design: each guide asks you to decide your limits before you collect. The saved calculation then compares each estimate with the limit you set, and only limits you enter or adopt count.

Every guide gives a suggested starting plan: an example design with its total, such as two materials measured 20 times each in one run. Adapt it to your own question; LabRuns never checks a study against it. The calculation minimum is the least data the software needs to calculate. At that minimum, the answer often has a wide interval.

A specimen measured five times is one specimen with five replicates, and the guides count both. The experimental unit is the thing the question is about: a specimen in a method comparison, a subject in a reference interval, a material in a bias check. Replicates of one unit share its matrix, its preparation and its run, so all they add is precision to that unit's mean. Counting them as separate units would make an interval look narrower than the evidence supports. How each study counts its units is in Counts and denominators, and why it matters for intervals is in About confidence intervals and independent results.

LabRuns calculates each estimate and compares it with the limits you entered. When a saved result meets its limits, your laboratory decides whether to adopt the method, start reporting or repeat the study. Signing off a report is optional; it records who reviewed the report and changes no outcome.