Conference on Data Quality for International Organisations, Newport, April Assessment of statistical data quality: The example of the Occupational Wages part of the ILO October Inquiry
Conference on Data Quality for International Organisations, Newport, April ILO October Inquiry Worldwide Survey of: Wages and hours of work occupations - 49 industry groups Retail prices - 93 food items Reference to October of each year
Conference on Data Quality for International Organisations, Newport, April Problems of data quality and comparability Unavoidable differences between: Concepts used Specifications of occupations and items Reference periods Types of sources Methods of data collection
Conference on Data Quality for International Organisations, Newport, April Steps taken to ensure data quality Detailed instructions for respondents Detailed descriptions of occupations and industrial activities according to ISCO and ISIC Detailed descriptions of food items including pricing units Questionnaire includes information on coverage, reference period, currency, source of the data etc. Pre-filled data for the previous year
Conference on Data Quality for International Organisations, Newport, April Quality assessment Regular checks for deviations and vague trends Reporting agencies are asked to verify Discrepancies due to methodological revisions, sample changes and different coverage Quality of data was assessed in 2001 as - excellent for 32 % - acceptable/good for 53 % - poor for 14 % - not acceptable for one data set
Conference on Data Quality for International Organisations, Newport, April Next steps Revision and expansion of the October Inquiry - changes in labour force structures - hourly wages/salary Improved collection of metadata Computerized validation checks in the database
ILO experience with gathering and disseminating meta-data on household income and expenditure statistics ILO Bureau of Statistics, Geneva Volume 6 of the ILO publications on Sources and Methods Volume 6 of the ILO publications on Sources and Methods
Conference on Data Quality for International Organisations, Newport, April Previous Practice Method 1: –Questionnaire with open response (in English, French, Spanish) or –Extract metadata from national publications, other documents –Then manual preparation of description, and translate into other two languages Method 2 –Referring previous descriptions back to countries for updating, then translate revision into other two languages
Conference on Data Quality for International Organisations, Newport, April Problems with previous method Countries were not always be sure of the level of detail to provide, requiring follow-up Extensive effort to prepare standardised description of metadata Extensive translation of results (about $50,000 each publication)
Conference on Data Quality for International Organisations, Newport, April New method developed in 2003 Questionnaire with tick boxes offering choice of answers (plus “Other, specify”) Tested and revised Translated into French and Spanish Computer programmes developed to: –Enter data into a database –Edit the input –Produce analyses of database (cross-tabulations) –Produce HTML output in three languages in standard format
Conference on Data Quality for International Organisations, Newport, April Advantages of new system Is believed to reduce reporting burden for countries –Tick boxes are faster and easier to complete –Gives guidance on the detail required in the response) Provides a database that can be analysed and cross- tabulated (not previously available) Reduces translation Computer development is re-useable to some extent –has already been adapted for a metadata system on –has already been adapted for a metadata system on statistics on the employment situation of persons with disabilities
Conference on Data Quality for International Organisations, Newport, April Disadvantages of new system Requires more care in questionnaire design and extensive knowledge of possible alternatives Requires more computer input