|
Activity Number:
|
153
- Developing Multi-Purpose Imputed or Synthetic Data for Official Statistics
|
|
Type:
|
Invited
|
|
Date/Time:
|
Monday, July 29, 2019 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Government Statistics Section
|
|
Abstract #300278
|
Presentation
|
|
Title:
|
Calibrated Imputation Under Edit Restrictions
|
|
Author(s):
|
Ton De Waal* and Jacco Daalmans
|
|
Companies:
|
Statistics Netherlands and Statistics Netherlands
|
|
Keywords:
|
imputation;
edit restrictions;
calibration;
item non-response
|
|
Abstract:
|
Institutes that collect statistical data on persons, households or enterprises are often confronted with the problem that items of otherwise responding units are missing in the observed datasets. The most common solution to this problem is to impute the missing data. The imputation problem is sometimes complicated owing to the existence of edit restrictions that have to be satisfied by the data. Examples of such edit restrictions are that a male cannot be pregnant and that the profit of an enterprise has to be equal to its turnover minus its costs. Records that fail the edit restrictions are inconsistent, and are hence considered incorrect. A further complication of the imputation problem is that counts or values of variables sometimes have to sum up to known or previously estimated totals. Standard imputation methods do not take such edit restrictions and known totals into account. In this talk we will present a method for imputation of missing categorical data that does take edit restrictions into account and that ensures that sums are calibrated to known or previously estimated totals.
|
Authors who are presenting talks have a * after their name.
