AN APPLICATION OF DATA EDITING METHODS TO IMPROVE DATA QUALITY OF SHORT-TERM BUSINESS STATISTICS

Improving statistical data quality is a key objective of the Romanian National Institute of Statistics (RNIS), and major efforts are devoted to produce and provide accurate data at the proper time for the proper users. It is, after all, the foundation of official statistics credibility. In the production process of short-term business statistics-STS, specialists have to carefully judge the trade-off between data quality and usefulness: More accurate data means time and resources, but it is no longer of interest for users if the results are released too late. Bearing in mind the tight requirements, the Romanian National Institute of Statistics (RNIS) undertook several actions to reform the production process of STS. One step was to redefine the entire data collection and capturing process; the second one was to modernize the data processing instruments. In this second step, SAS (R) Enterprise Guide (R) is heavily used in several departments for data editing, imputation, grossing-up, and tabulation, in order to reduce the time till the release of official results and to secure their overall quality. The paper gives a picture ofthe SAS (R) applications usedinRNIS, implementing the recommended methods for data control and editing, applied in the area of shortterm business surveys. Some considerations upon one method broadly used in the data editing process are presented, basically required for item-non-response treatment by means of classic imputation methods-mean value, hot-deck, historical data and, more seldom, cold-deck imputation-ensuring, also, the data-editing controls.