SELEKT – A Generic Tool for Selective Editing

Abstract The aim of selective editing is to make the often resource-demanding traditional editing process in business surveys more effective without a substantial loss in the precision of the output statistics. Recently, Statistics Sweden has developed a generic software package for selective editing called SELEKT. The method underpinning the software promotes continuous measurement of the suspicion of error response rather than a dichotomous measure using traditional edits. SELEKT is flexible and can be used in the production of surveys with different designs. Business surveys have diverse output regarding the number of variables, statistical measures, and domains of study definitions. The key objective of selective editing is to rank suspected errors in data according to the anticipated impact on the output. The software therefore has options to set different weights for different parts of the output to meet the needs of the main users of the statistics. Statistics Sweden has implemented SELEKT in eleven surveys to date. The experience gained will be used to provide recommendations on how to perform selective editing. This article will give an insight into SELEKT and its underlying theoretical base.