With low quality data, any statistical analysis may be meaningless. Data editing is a process to identify and correct potential respondent errors to improve survey data quality. Yet, data editing is resource-intensive process. Selective editing methods seek to identify the most influential response errors for manually replacing correct values instead of editing data in every detail. This paper compares three selective editing methods, Hidiroglou-Berthelot (HB), Score Function (SF), and Robust Regression (RR), for periodic continuous survey data. These three methods require no assumed models for the data. Simulation studies and real data analyses were conducted to assess their efficiencies in reducing the bias of the data and their type I error rates when the numbers of identified errors is limited.