Journal of Official Statistics, Vol.7, No.2, 1991. pp. 137–152
Tolerance Groupings for Editing Banking Deposits Data: An Analysis of Variance of Variances
David A. Pierce and Laura L. Bauer-Gillis
Abstract:This paper presents an application of analysis of variance techniques which has led to a substantial simplification of editing criteria for U.S. Federal Reserve System data. The application involves comparing measures of spread of the dollar and percentage changes in the data, on which the editing tolerances are based, and using multiple comparison methods to classify the financial institutions which report these data into homogeneous groups for constructing the tolerances. The data display extensive non-normality and cell heteroskedasticity, and methodology is developed to deal with these problems. The classification of institutions resulting from this study, based on size, location, and type, has resulted in more than an 80% reduction in the number of edit tolerance groupings relative to those previously in use, and an increase in efficiency and accuracy of the editing process.
Keywords:Data editing; numerical editing; quality control; U.S. money supply; ANOVA; outlier detection.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group