|
Activity Number:
|
588
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Thursday, August 6, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Government Statistics
|
| Abstract - #303914 |
|
Title:
|
Generating Multiply Imputed Synthetic Data Sets for a German Establishment Survey: The Agency's Perspective
|
|
Author(s):
|
Joerg Drechsler*+
|
|
Companies:
|
Institute for Employment Research
|
|
Address:
|
Regensburger Strasse 104, Nuremberg, International, 90478, Germany
|
|
Keywords:
|
synthetic data ; multiple imputation ; establishment survey
|
|
Abstract:
|
In 2006 the German Institute for Employment Research (IAB) launched a research project to investigate the chances of releasing a scientific use file of its establishment survey protected by means of generating multiply imputed synthetic data sets. After several pre-tests and simulations a release of the last wave of the survey for research purposes is planned for early summer 2009. In this talk we will discuss some of the challenges of implementing this approach for a real data set, where the data is anything but normal, missing values and skip patterns complicate the modeling, logical constraints have to be considered and a disclosure review board has to be convinced that a release is save without being too technical when describing the methodology.
|
