How to Obtain Additive Estimates of Missing Cell Values in Tabular Data Containing Non-Additive Rounded Cells

• Activity Number: 462
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Government Statistics Section
• Author(s): Ramesh Dandekar*+
• Companies: Energy Information Administration
• Keywords: missing value analysis ; disclosure ; mathematical programming ; tabular data ; Maximum Likelyhood estimate ; L1 norm regression

Abstract

Statistical agencies routinely use complementary cell suppression procedures to protect sensitive cells in tabular data. In addition to the cell suppression procedures, often published tabular cells are also rounded to a certain base value (a multiple of base 10 is common). Deriving additive estimates for suppressed cell values, which corrects for error introduced by the rounding procedure, creates a special challenging situation. By using mathematical programming techniques, in this paper we demonstrate how to simultaneously eliminate the rounding error and obtain additive estimates for suppressed cells so that resultant table is fully populated and has complete additive properties. The technique could be used for other statistical applications such as: a time series of data balancing (e.g. 12 months to annual totals), or a respondent level combined data editing/imputation procedure