Messages come in three strengths:
- Cautions keep the result and say what limits it. They appear as study notes.
- Missing estimates show Not estimated, with a reason.
- Stopped calculations save nothing new. See calculating results and analysis problems.
Where messages appear#
| Where | What it shows |
|---|---|
| Red banner at the top of the page | A stopped calculation or an error |
| Yellow banner at the top of the study | Results out of date, or limits changed after the calculation |
| Notes below the result | Data issues and why an estimate is missing |
| Why panel and acceptance table | The reason for each Undecided, Not evaluated or Not applicable outcome |
| Tables of estimates | Not estimated in place of a value |
| Report page 1 | A marker such as *1 where a note applies; the note is under Notes to page 1 in 2 Study notes |
| Report, Results tables | The reason each limit is not met or undecided, as a footnote |
| Report, 2 Study notes | Every caution, numbered and grouped by level |
Types of study note#
| Kind | Example | What to do |
|---|---|---|
| Design smaller than recommended | 20 specimens compared where at least 40 are recommended | Treat as preliminary, or add specimens and recalculate |
| Meaning of the comparative method | Positive and negative agreement with a method that is not a reference standard | Report them as agreement, not as sensitivity or specificity |
| How rows were counted | Blank rows ignored in LoQ mode | Check each row's role |
| Claim beyond the data | A claimed concentration above the highest one studied | Study the claimed concentration, or narrow the claim |
| Claim not supported | An 8 h room-temperature stability claim the data do not support | Shorten the claim, or collect more evidence |
| Special values | A left-censored blank result ranked below every exact result | Nothing. The note says how it was handled. |
| Limit not used | A limit set for a level that isn't in the data | Correct the level and recalculate |
| Numeric convention | Rows imported with a decimal point while the study uses a decimal comma | Check those rows against the file. See Numbers, units and identifiers. |
| Limits changed | Limits changed after an earlier calculation | Nothing. The note records the change. See Acceptance limits. |
Estimates not calculated#
A missing estimate shows Not estimated on screen and "—" with a footnote in the report. Common reasons:
| Estimate | Reason |
|---|---|
| CV | Fewer than 2 results, a mean at or near zero, or negative results |
| Recovery at a blank level | The expected concentration is zero |
| A confidence interval | See when an interval can't be calculated |
A limit that needs a missing estimate is Undecided, so the study is at best Undecided. See Criterion outcomes.
What to do with a note#
- Read the whole message. It names the group, row or setting.
- Decide whether it changes your conclusion. A small sample may be fine for a preliminary study but not a final one.
- If the data or setup are wrong, correct the value (or exclude the row with a reason), save and recalculate.
If the stored report or a PDF can't be opened or made, see report problems. See also Analysis problems.