|
Activity Number:
|
365
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #304439 |
|
Title:
|
Identifying Outliers When Creating an Imputation Base for the Quarterly Financial Report
|
|
Author(s):
|
Melvin J. McCullough*+ and Terry L. Pennington
|
|
Companies:
|
U.S. Census Bureau and U.S. Census Bureau
|
|
Address:
|
4600 Silver Hill Road, Suitland, MD, 20746,
|
|
Keywords:
|
imputation ; outlier
|
|
Abstract:
|
The Quarterly Financial Report Survey (QFR) collects income and balance sheet data for most manufacturing corporations and for large mining, wholesale, and retail corporations. Unit non-respondents are imputed using a combination of ratio and mean imputation. In order to enhance the imputation process by eliminating influential cases from the base, we investigated an iterative regression approach of outlier detection. The approach utilizes a combination of two regression diagnostics, leverage and studentized residuals. We compared the effectiveness of the "regression fits" approach to the Hidiriglou-Berthelot method of outlier detection for several positive valued QFR items. To evaluate the effectiveness of the approaches, we created plots of inliers and outliers. The "regression fits" approach can also detect outliers for negative valued QFR items.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2009 program |
