Research concepts
Missing data and detection limits in a table
Missing data and detection limits: Does every empty cell or symbol have an explicit reason?
Source editorial review:
Before interpreting the result
Does every empty cell or symbol have an explicit reason?
- Separate not measured, not detected, below limit and excluded data.
- Keep units and applicable limits for each column.
- Review how the analysis treated incomplete or censored values.
Missing data and detection limits
An empty cell, an unmeasured value and a result below a limit represent different situations. Replacing all with zero changes the table's claims and can bias summaries. Before calculating, classify why each value is absent and what partial information the analytical procedure preserves.
A reading case: what to check
In a batch comparison, keep explicit codes for not analyzed, not detected and not quantifiable, with definitions. If statistical handling is needed, document the rule and assess suitability rather than hiding substitutions in a spreadsheet. Readers should distinguish observations from decisions made to analyze them.
What to preserve in the record
Filling cells with zero manufactures precision the table does not contain. The reason for absence is part of the data and can change interpretation of the dataset.
Questions and answers
Does every empty cell or symbol have an explicit reason?
An empty cell, an unmeasured value and a result below a limit represent different situations. Replacing all with zero changes the table's claims and can bias summaries. Before calculating, classify why each value is absent and what partial information the analytical procedure preserves.
What should the review record preserve?
Filling cells with zero manufactures precision the table does not contain. The reason for absence is part of the data and can change interpretation of the dataset.