|
Activity Number:
|
468
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Thursday, August 7, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Government Statistics
|
| Abstract - #300855 |
|
Title:
|
Truncated Triangular Distribution for Multiplicative Noise and Domain Estimation
|
|
Author(s):
|
Jay J. Kim*+ and Dong M. Jeong
|
|
Companies:
|
National Center for Health Statistics and Korea National Statistical Office
|
|
Address:
|
3311 Toledo Road, Hyattsville, MD, 20782,
|
|
Keywords:
|
Masking ; confidentiality ; truncation ; noise generation
|
|
Abstract:
|
The truncated triangular distribution has been used for masking microdata. The random variable following the truncated triangular distribution can serve as a multiplicative noise factor for the masking. The most desirable candidate distribution is the symmetric one which is centered at and truncated symmetrically about 1. This is because the multiplicative noise factor of 1 or very close to 1 does not protect confidentiality at all. The probability density function of the truncated triangular distribution has been developed by Kim [4] and applied to the 2006 Korean Householder Income and Expenditure Survey (HIES) data. Formulas for the domain estimation for the data masked by the multiplicative noise mentioned above are developed. In this paper, we will show domain estimation formulas and some results of the application of the truncated triangular distribution on the HIES data.
|
