|
Activity Number:
|
35
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 29, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #309309 |
|
Title:
|
Weight Adjustments for Fractional Regression Hot Deck Imputation
|
|
Author(s):
|
Min Hui Paik*+ and Michael D. Larsen
|
|
Companies:
|
Iowa State University and Iowa State University
|
|
Address:
|
204 Snedecor, Ames, IA, 50011,
|
|
Keywords:
|
missing data ; fractional imputation ; jackkinfe variance estimation
|
|
Abstract:
|
Fractional regression hot deck imputation (FRHDI), suggested by J. K. Kim, imputes multiple values for each missing dependent variable. The imputed values are equal to the predicted value based on the full observed cases plus multiple random residuals chosen from the set of empirical residuals. Fractional weights are chosen to enable variance estimation and to preserve the correlation among variables. The FRHDI method can be viewed as a special case of fractional hot deck imputation (FHDI). In some circumstances with some starting weight values, existing procedures for computing FHDI weights can produce negative values. We discuss procedures for constructing nonnegative adjusted fractional weights for FRHDI.
|
