Exploratory Data Analysis of Economic Census Products: Methods and Results

• Activity Number: 409
• Type: Topic Contributed
• Date/Time: Tuesday, August 11, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Yukiko Ellis* and Katherine Jenny Thompson
• Companies: U.S. Census Bureau and U.S. Census Bureau
• Keywords: Economic Census ; Missing Data ; Exploratory Data Analysis ; Response Propensity Models

Abstract

Beginning in 2017, the Economic Census will begin using the North American Product Classification System (NAPCS) to produce economy-wide product tabulations. This marks a major departure from the current collection method that explicitly links products to industry. Motivated by this collection change, the U.S. Census Bureau conducted a study to investigate methods of treating missing product data in the Economic Census, with the goal of recommending a single imputation method to produce product data in all trade areas that is statistically defensible and operationally practical. The validity of an imputation method is highly dependent on the nature of both the reported data and on the nature of missing data (e.g., factors that contribute to response). This paper presents an exploratory data analysis of empirical data from selected industries with common products under NAPCS at the national industry level that explores these factors, describing the methods and presenting the results. The collective results were used to recommend candidate imputation methods, to develop imputation cells, and to inform the subsequent evaluation study by providing realistic response propensity models.