What is Quality?
”Satisfy the customers needs and expectations at a
competitive price” (Deming)
Brief: ”Conformance to requirements” or
”fitness for use”
Quality and metadata
The users define quality
What is quality in statistics?
– Relevance and completeness
– Timeliness and punctuality
– Accuracy
– Comparability and coherence
– Accessibility and clarity
Improvement requires improvements in processes
Documentation (metadata) is necessary, both for the users and
the producers
Total quality
Process quality
Product quality
User needs
“User needs” are the point of departure for systematic quality work and for determining quality indicators.
The users demand “product quality” which encompasses desired attributes of the product.Costs must be taken into account.
Study of processes is a precondition for improvement. This includes the identification and measurement of key process variables affecting product quality and costs.
Types of metadata (for quality management)
Documentation for the users of statistics
– Understand and use statistics correctly
– Overview and navigation
– Consider quality
Documentation for data providers
– Information needed to provide correct data Process documentation for producers
– Information to control and improve processes– Current Best Methods and benchmarking
Quality information (overlapping information for users and producers)
Metadata = (Structured) documentation
Examples
Information for users: About the statistics etc.
Process information
Quality information: Products and processes
(overlapping information above)
Quality information – about products
EXAMPLE: Timeliness
Number of weeks from reference period to release of statistics
Number 1999 2000 2001 2002 Target Result Target Result
Monthly statistics 12 3,8 4,1 3,8 3,9 3,8 3,8 Quarterly statistics
14 8,9 8,4 8,9 8,8 8,8 8,8
Annual statistics 82 48,5 46,9 43,3 42,0 42,0 43,1
Quality information - about processes
EXAMPLE: Response rates in Statistics Norway
Surveys Number 2000 2001 2002 Target Result Target Result Target Result
Mandatory surveys
- Wage statistics 11 100 97 100 97 100 96 - Others 17 96 94 95 94 97 95 Voluntary surveys 18-21 78 72 77 73 79 77
Conclusions
Many different approaches and metadata exist, user friendliness vary
Different needs (of external and internal users) have to be the point of departure for metadata approaches
The different needs can be taken care of by different metadata systems, or by different levels of (the same type of) metadata for different users
Systematic approaches should be promoted