A variable is not just a heading. It is a promise that the same characteristic was defined and recorded in a comparable way for each relevant unit.
Juniper defines days_to_settlement as whole calendar days from invoice issue
through application of the final settling payment. Changing the endpoint to
the first payment date or bank-deposit date creates a different variable, even
if the column keeps the old name.
Read the dictionary before the values
A usable definition identifies:
- the observational unit and eligible records;
- what event starts and ends the measurement;
- unit, precision, and treatment of partial or missing values;
- allowed categories and whether their order matters;
- source system, period, cutoff, and transformation; and
- whether the field is an identifier, explanatory descriptor, response, or derived result.
The distinctions govern the analysis. Invoice ID is nominal text even though its characters include digits. Review priority is ordered, but the gap between “routine” and “elevated” is not a measured distance. Settlement days have a meaningful zero and equal calendar-day increments.
Derived variables inherit assumptions
Ratios, returns, common-size percentages, and days measures are variables too. Their definitions carry numerator, denominator, timing, sign, and rounding choices. A derived field does not become comparable merely because software calculated it consistently. Trace it back to the source fields and the business events those fields represent.
Put the concept to work
Understand this concept
- Explain a variable through its case, definition, allowed values, units, coding, period, and analytical role rather than through its column name alone.
Apply this concept
- Build and apply a compact data dictionary that distinguishes identifiers, categorical descriptors, quantitative responses, dates, and derived fields.
Learning resources
Choose a lesson, try an application, or inspect the sources behind this concept.
Build on these ideas
- Observational unit — Understand
To understand this concept: Required. A variable is recorded about a defined observational unit.
- Variable — Understand
To apply this concept: Required. A data dictionary operationalizes the variable definition and coding contract.
Lessons
Worked examples and cases
Practice
Common mistaken ideas
Sources
Related concepts
Show 2 more related concepts
Use this idea next
- Distribution — Understand
Required level here: understand. Required. A distribution describes the observed or possible values of one defined variable.
- Measurement scale — Understand
Required level here: understand. Required. A measurement scale describes the structure of a defined variable.
- Nonsampling error — Understand
Required level here: understand. Helpful. Measurement error concerns how a defined characteristic was recorded.
Show 3 more next steps
- Parameter — Understand
Required level here: understand. Required. The parameter summarizes a defined characteristic.
- Population — Understand
Required level here: understand. Helpful. The question must name which characteristic is sought for the population.
- Variable — Apply
Required level here: understand. Required. A data dictionary operationalizes the variable definition and coding contract.