Wang & Strong (1996) - Beyond Accuracy: What Data Quality Means to Data Consumers
Key Insights
- Wang and Strong develop the first comprehensive data quality framework, identifying 15 dimensions in four categories (intrinsic, contextual, representational, accessibility), changing how organizations think about data quality beyond accuracy.
Extracted Variables
No quantitative variables detected.
Study flow (PRISMA-style)
Machine-extracted from the synthesis; review before citing.
Edit on GitHub — registry.json
Background
"Data quality" used to be a synonym for accuracy: fix the bad values and the data is good. Wang and Strong's study demolished that assumption empirically. Working with data consumers in organizations, they asked what quality actually means to the people using data — and derived a framework that still defines the field.
The Framework
Through interviews and sorting studies, the authors identified 15 data quality dimensions in four categories. Intrinsic quality is about the data in itself: accuracy, objectivity, believability, reputation. Contextual quality depends on the use case: relevance, value-added, timeliness, completeness, appropriate amount. Representational quality concerns the data's form: interpretability, ease of understanding, representational consistency, concise representation. Accessibility quality covers the ability to obtain and protect it: accessibility and access security.
Deep Dive
The key finding is that consumers judge quality across all four categories simultaneously — a dataset can be perfectly accurate yet unusable because it is late, irrelevant, or locked down. The paper's methodology (grounded theory with real consumers) is as influential as its results: it legitimized studying quality from the consumer's perspective instead of the producer's. The framework became the basis of the MIT Total Data Quality Management program, the Data Quality Assessment Framework (DQA), and the vocabulary behind ISO 8000 data-quality standards.
Why It Matters
Every modern data-quality dashboard — completeness, timeliness, uniqueness, validity — is a simplification of Wang and Strong's dimensions. For engineering teams the framework is a checklist: measuring accuracy alone will miss the reasons users stop trusting the platform.
Key Takeaways
- Quality is multi-dimensional and context-dependent; "good data" has no definition without a use case.
- Instrument the four categories: intrinsic (accuracy), contextual (timeliness, completeness), representational (consistency), accessibility (security, availability).
- Consumer surveys beat producer intuitions for choosing which dimensions matter most.