Field note
When should a data dictionary include units?
Every measure should name its unit or explicitly say unitless. If measurement height, basis, precision, denominator, or currency matters, record it rather than relying on a short field name.
State the unit explicitly
Give every measure a unit, and where a field genuinely has none, write unitless rather than leaving the box empty. An empty unit reads as work left undone; the word unitless reads as a decision somebody made. The review raises a warning on any field with the measure role and no unit.
Include the measurement basis
The unit alone is often not enough. Record where and how the quantity was measured, at what precision, and against which denominator or base: the sample's trunk circumference is measured around the trunk at the survey team's stated height, and without that the number is ambiguous between two common measurements. Currency fields need the currency, and where it matters, the date basis of any conversion.
Do not infer units from values
The studio accepts whatever unit text you type. It does not convert, normalise, validate or compare units against the values, and it cannot notice that a column named in centimetres holds millimetres. Reading a scale off the field name or off the range of the values is exactly the inference that produces silent errors when two datasets are combined.
Worked example from the sample project
The sample declares trunk_cm as a measure with the unit centimetres, and the definition adds the measuring basis and height. Strip the unit out and the column becomes a number whose scale a reader has to infer from the field name and the values, which is exactly the inference that produces silent errors when two datasets are combined.
The other four documented fields are correctly unitless, and saying so explicitly is worth doing. A blank unit box reads as work left undone; an explicit statement that the field is unitless reads as a decision.
What this does not do
- The studio does not convert units, normalise them, or check them against the values.
- It has no unit vocabulary and accepts whatever text you enter.
- It cannot detect that a column named in one unit actually holds another.
Do this in the studio
Name the unit or say unitless; a blank box is not an answer.
- Open the Field folios station and set the unit on every measure.
- Record the measurement basis, precision or denominator in the definition when it matters.
- Say unitless explicitly on counts, codes and identifiers.
Open the studio at the Field folios station