STATISTICS Common mistake
Treating repeatability as routine precision
Reporting a CV without the design
Confusing the
ANOVA P value with acceptability
Adding CVs Testing only one
Confusing precision with bias
Why it is a problem
Replicate measurements in a single run provide a narrow estimate. They cannot estimate between-day or between-run variation unless those sources are included in the design.
A CV alone does not explain how the estimate was produced or what conditions it represents.
The P value may show that a between-group component is detectable, but it does not decide whether the imprecision is acceptable.
Variance components are combined on the variance scale. CVs and SDs are not additive.
Imprecision may vary across the measuring interval. Low concentrations, abnormal ranges and decision limits may behave differently from the normal range.
Precision describes scater between repeated measurements.
Bias describes systematic difference from a target, reference value or comparator method. A method can be tightly grouped but still wrong.
Table 5. Common mistakes in interpreting precision studies. The calculation is not the end of a precision
study. The result still needs to be interpreted in relation to the study design, the
concentration tested and the intended use of the measurement procedure
a non-significant F-test may still be unacceptable if repeatability is poor. Interpretation should therefore focus on the estimated components. Is most variation occurring within groups, suggesting poor short-term repeatability? Or is most variation occurring between groups, suggesting day-to-day, run-to- run, lot-to-lot or analyser related effects? The answer depends on what the groups represent in the design. The result should be reported in a
way that preserves the design. A weak statement is: ‘the CV was 5.13%’. A better statement is: ‘within laboratory imprecision was estimated as 5.13% at approximately 33 units/L in a five-group, five-replicate precision study’. This tells the reader what was estimated and what evidence produced the estimate. A short verification study provides
useful local evidence, but the estimates are still based on a limited number of observations over a limited period. They
should be interpreted as estimates of performance, not permanent properties of the method.
Common mistakes, limitations and points of caution Precision studies are easy to misinterpret when the numerical output is separated from the study design. Table 5 summarises common mistakes and how they can be avoided when reporting and interpreting precision data.
Practical take-home points A precision study is a variance decomposition exercise, not just a CV calculation. The study design determines which sources of variation can be estimated. Repeatability, within-laboratory precision and reproducibility answer different questions. ANOVA separates within-group
20
WWW.PATHOLOGYINPRACTICE.COM August 2026
and between-group variation in a structured precision study. Variances add; CVs do not. The F-test helps identify whether between-group variation is larger than expected from replicate scatter, but it does not decide clinical acceptability. Precision estimates should be reported with their material, concentration, design and type of precision estimated.
PPi
Key reading to support the article Clinical and Laboratory Standards Institute.
EP05 Plus: Evaluation of Precision of Quantitative Measurement Procedures. Clinical and Laboratory Standards Institute. EP15-A3: User Verification of Precision and Estimation of Bias. International Vocabulary of Metrology – Basic and General Concepts and Associated Terms (VIM). 3rd edn. JCGM 200:2012. International Organization for Standardization. ISO 5725 series. Accuracy (trueness and precision) of measurement methods and results. Chesher D. Evaluating assay precision. Clin Biochem Rev. 2008 Aug;29 Suppl 1(Suppl 1):S23-6.
Dr Stephen MacDonald is Consultant Clinical Scientist, The Specialist Haemostasis Unit, Cambridge University Hospitals NHS Foundation Trust, Cambridge Biomedical Campus, Hills Road, Cambridge CB2 0QQ.
+44 (0)1223 216746 Beter approach State clearly whether the estimate describes
repeatability, within-run precision, or within-laboratory precision.
Report the material tested, approximate concentration,
number of observations, number of groups, time period and type of precision estimated.
Interpret the P value as part of variance component analysis. Judge acceptability against intended use and predefined performance expectations.
Combine the relevant variances first, then calculate the SD and CV from the combined variance.
Test concentration levels that reflect clinically important parts of the measuring interval.
Report and interpret precision and bias separately.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56
