Repeatability

Run this study#

Your question: how much does the result vary when I measure the same material in one run?

Bring: two stable materials, one near a low and one near a high decision concentration, with enough volume for 20 measurements each, 40 results in total. This suggested starting plan follows Westgard's replication guidance. Add a level for any other important concentration.

1. Set your limit before measuring. Choose New study, enter the Analyte and Unit, pick Repeatability under Precision and choose Create study. On Set up, under Acceptance limits, choose Enter your own limit and enter the largest acceptable Repeatability SD (in the analyte unit) or Repeatability CV (in %). SD usually changes with concentration, so set one SD limit per level, or one CV limit for all levels.
2. Run the materials. Prepare them as instructed, check that calibration and QC are acceptable, then measure each material 20 times in one run on one day. Keep the readings in measurement order.
3. Enter the results. Record material ID, level, result, day, run and replicate, with one material ID per level and replicates numbered 1–20. On Data, choose Import a file and choose your file (start from Blank template (CSV) or repeatability.csv, replacing all its rows, including the excluded demonstration row). Answer any question the dialog asks, then choose Import into this study.
4. Calculate and review. Choose Calculate results; Results opens. Check that each level has 20 included results. Compare the repeatability SD or CV with your limit and look at the Measurements plot for drift. Investigate unusual readings before any correction or exclusion.
5. Print the report. On Report, choose Download PDF or Print.

What you get: within-run precision at the levels tested. It leaves out calibration and day-to-day effects, so it understates your laboratory's total imprecision: use multi-day precision for that, and note that a within-run CV used in a sigma calculation overstates sigma. Calculation minimum: two replicates per level.

Purpose#

For each level the study reports the mean, repeatability SD and CV, with a confidence interval for the mean. See repeatability in the glossary.

When to use it#

Use it for a quick within-run precision check, for example 20 replicates of a control after maintenance.

Use a different study when:

• replicates span several days or runs: Multi-day precision.
• you want the bias of the same replicates against a known value: Bias against an assigned value. Both studies can use the same file.

Study setup#

Several levels, for example a low and a high control, can share one study. Each level is calculated separately and needs one material, one day and one run.

Entering data#

On Data, choose Import a file or Add row. See Importing a file. A row you add by hand starts with level 1, day 1, run 1 and its row number as the replicate.

Correct a censored value such as <5 or exclude the row with a reason. See Missing and censored values.

The calculation stops when:

• a level mixes materials, days or runs. Split mixed levels into separate levels, or use a descriptive summary.
• a replicate number repeats within a level.
• a row is marked: a blank or non-numeric result, a missing ID, level, day or run, a negative result while Permit negative results under Calculation settings is off, or an exclusion without a reason.
• the analyte or unit is blank.

Then choose Calculate results. See Calculating results.

Statistics#

For each level:

Acceptance limits for this study#

• Repeatability SD and Repeatability CV limits compare each level's own estimate using the entered decimal results before display rounding. Equality meets an inclusive limit. A limit needs at least 2 included replicates in the level; with 1 it is Undecided.
• A CV limit needs a ratio scale (Measurement scale under Calculation settings, the default). On an interval scale, with no meaningful zero, use an SD limit in the measurement unit: a CV limit there is Undecided.

One unmet limit fails the whole study. See Acceptance limits, including More statistics, and Limit outcomes.

Worked example#

The study Glucose · single-run repeatability in the example project uses repeatability.csv: 21 rows of material QC-001 at level 1, day 1, run 1. One row, result 1004 with replicate E1, is excluded as a transcription error.

Settings: analyte Glucose, unit mg/dL, 95% confidence. Two limits:

• Within-run CV: Sample CV ≤ 0.8 %, set strict so that it fails.
• Mean near assigned value: Mean within 2 % of an assigned value of 100 mg/dL.

What the saved result shows#

The study status, and page 1 of the report, read Criteria not met because of the within-run CV.

Reading the results#

• Why: each limit with its observed value and outcome. Each line opens that limit on Set up.
• Results and acceptance: the limits, then mean, SD and CV. An estimate with a limit appears in the acceptance table instead. The count is in the verdict line.
• Measurements: replicates in run order (or Histogram) with the mean as a dashed line. A steady trend within the run points to drift.
• Distribution diagnostics: Q–Q plot, box and strip plot of every point.
• Additional statistics and calculation history: median, variance, quartiles and the mean CI.

When a limit is not met#

• Recheck data entry for typos, a column assigned to the wrong field or the wrong decimal separator. One stray value, like the excluded 1004, shifts the mean and SD a lot.
• If the spread is real, that is the finding. Check pipetting, reagent age, temperature drift or the control material, fix the cause and re-run.

Statistical detail#

Sources: NIST/SEMATECH e-Handbook §1.3.5.6 and §1.3.5.2. See Methods and sources.