False Discovery Rate in Large-Scale Data Error Localization : Assessing Survey Editing Methods' Performance

Statistical data editing means identifying potential response errors in the data. Without regard tohow flagged observations will be modified, data editing is subject to two types of errors: labeling acorrect observation as erroneous and not identifying an incorrect value. There is no statisticalcriterion to decide how many observations should be edited. Over-editing can increase data errors,degrade data quality, change the data structure and increase costs. Error localization consists of aseparate test on each observation, where the null hypothesis states that the observation is error freeand the alternative states that the observation is erroneous. The False Discovery Rate (FDR) is thefraction of false-positive findings among those deemed to be erroneous. Because FDR control isrelated to the number of edited observations, imposing an FDR requirement specifies the numberof outliers to be edited, thereby controlling overediting. In this presentation, we experiment with alocal FDR method on error localization and evaluate its performance and the performance of well-established editing methods on simulated data.