How should I store a measurement reported as “below the detection limit”?

Store a bound, not a fake exact number. In Busabase, split “below the detection limit” into a qualifier, a limit value, and an optional numeric value — so later averages and charts do not quietly treat every “<0.01” as 0.01 or zero.

*Diagram · Qualifier + limit + value*

Recommended fields (one Measurement row)

Optional flags: IsCensored (checkbox) when qualifier ≠ =, to filter charts quickly.

How to use it

1. Lab says “less than 0.01” → Qualifier=<, Limit=0.01, leave Value empty (or leave Value unused in formulas).
2. Exact reading 0.04 → Qualifier==, Value=0.04, Limit can still store the method LOD for reference.
3. Views / exports for people: show Display (<0.01), not a coerced number.
4. Calculations: only average rows with Qualifier==, or apply an explicit censored-data rule your science team agrees on — do not invent one silently in a formula named “Value”.

How to verify

• A “<0.01” row never appears as 0.01 or 0 in a number-only column used by rollups.
• Sorting / filtering by Qualifier can isolate censored vs exact results.
• Someone opening the record can see both the limit and that it is a bound.

Boundaries

• Substituting LOD/2 or 0 into Value for “convenience” will bias every chart that uses that field — if you must impute, store imputed values in a separate field and label them.
• Busabase number fields do not natively mean “less than”; the qualifier is what carries that meaning.
• If an Agent or import writes measurements, instruct it to map “<x” / “LOD” / “ND” into Qualifier+Limit, not into Value alone.

If you paste one anonymized lab line format you receive, we can suggest an import mapping into these fields.