Data Quality Dimensions
Also known as: dimensions of data quality.
Named aspects — accuracy, completeness, consistency, timeliness, uniqueness — that turn the vague idea of clean data into checks you can measure.
Synthetic dataExamples on this page use synthetic data created for illustration.
Definition
Data-quality dimensions are named aspects along which data can be good or bad, such as accuracy, completeness, consistency, timeliness and uniqueness. They turn the vague idea of clean data into checks that can be measured and reported separately.
In plain terms
“The data is bad” is not actionable. “8% of prices are missing and the newest row is three days old” is. Each dimension asks one question: is it right (accuracy), is it all there (completeness), does it agree with itself and other sources (consistency), is it recent enough (timeliness), is each thing recorded once (uniqueness), does it follow the rules of its format (validity).
Dimensions are independent: a file can be perfectly unique and badly out of date. Reporting them separately shows what to fix, and which of them matters depends on the use.
Example
A synthetic file of 1,000 price rows:
| Dimension | Check | Result |
|---|---|---|
| Completeness | Rows with a price | 920 of 1,000 = 92% |
| Uniqueness | Distinct item ids | 985 of 1,000 = 98.5% (15 repeats) |
| Validity | Prices greater than zero | 990 of 1,000 = 99% |
| Timeliness | Age of the newest row against a 1-day promise | 3 days: fails |
One overall “quality score” would hide that the real problem is freshness and missing prices.
Common misunderstandings
- “Quality is a single number.” Averaging dimensions hides which one failed and whether it matters for the use.
- “The goal is 100% everywhere.” Fit for purpose is the goal; perfect completeness may cost more than the decision is worth.
- “There is one official list.” Authors use different lists and names; what matters is that yours are defined and measured.
References
- Richard Y. Wang, Diane M. Strong. Beyond Accuracy: What Data Quality Means to Data Consumers . Journal of Management Information Systems, 1996. DOI: 10.1080/07421222.1996.11518099
- Carlo Batini, Cinzia Cappiello, Chiara Francalanci, Andrea Maurino. Methodologies for Data Quality Assessment and Improvement . ACM Computing Surveys, 2009. DOI: 10.1145/1541880.1541883