Contact

GlossaryPublic records, privacy, and data quality

Data quality

Data quality describes whether information is suitable for a particular use. Learn how accuracy, completeness, consistency, and timing affect public reports.

Data quality describes how well information supports its intended use. A useful assessment considers whether the records are accurate, sufficiently complete, consistently defined, and timely enough for the question being asked. Quality is not established by an official source label, a large row count, or a polished chart alone.

Quality depends on the question

The UK Government Data Quality Framework defines quality in relation to purpose. The U.S. Geological Survey recommends reviewing data during the project, maintaining documentation, and tracking changes. Together, these principles support a practical question: is this information suitable for the conclusion being drawn? Government framework, USGS assessment guidance.

A historical file can be valuable for a long-term workload study even if it is unsuitable for describing today's activity. A recent file can be useful for identifying preliminary patterns while still awaiting corrections. Neither assessment is meaningful without the proposed use.

Example

Hypothetical: A department publishes 12 months of incident counts. The file includes every month and uses consistent incident identifiers, but it contains no dispatch or arrival timestamps.

The file may support a monthly count comparison after its category and coverage rules are checked. It cannot establish average response time. That limitation concerns the information needed for the second question; it does not automatically make the incident counts unreliable.

Common confusion

Completeness is one part of quality. A table can have no blanks and still contain incorrect categories. Likewise, valid formatting does not establish factual accuracy: a correctly formatted date can be the wrong date.

Freshness is another separate concern. A recently updated page may contain older records. The page's update time and the period represented by the data need to be read separately.

A single quality score can hide these distinctions. If a publisher provides a score, its components and weighting should be available. Readers need to know whether the score reflects missing fields, duplicate records, source coverage, or something else.

What makes an assessment useful

Look for specific findings: which fields were checked, which records were excluded, what remains uncertain, and how those limitations affect the result. A note such as “arrival times are missing for 8% of eligible responses” is more informative than an unexplained claim of high quality.

Documentation also helps distinguish a source limitation from a later processing error. Clear definitions, transformation notes, and version dates let readers trace how the published measure was produced and judge whether it answers their question.

Sources