Install the app
Install this application on your home screen for quick and easy access when you’re on the go.
Just tap then “Add to Home Screen”
Install this application on your home screen for quick and easy access when you’re on the go.
Just tap then “Add to Home Screen”
Menu
Install this application on your home screen for quick and easy access when you’re on the go.
Just tap then “Add to Home Screen”
The course aims to give a fundamental understanding of methods and techniques to handle missing data in general.
This course focusses on multiple imputation (Rubin 1987) as a general method to analyze incomplete data sets. By multiply imputing missing data in general or data missing due to item nonresponse in particular, valid statistical inference can be achieved using standard statistical software packages. Multiple imputation (MI) is based on a Bayesian framework, therefore some basic knowledge of Bayesian inference is required.
Apart from MI itself we will also cover the theoretical foundations of incomplete data analysis and introduce the participants to underlying assumptions such as MAR or ignorability. In order to appreciate the benefits of MI, we will also give an overview of inadequate ways of handling missing data, and describe their respective shortcomings. Then we will introduce the multiple imputation framework and provide some insights to the multiple imputation theory.
Susanne Rässler (PhD in statistics, Habilitation in statistics and econometrics) is full Professor at Otto-Friedrich-University Bamberg, where she leads the Department of Statistics and Econometrics.
Prior to that, she was head of the Competence Centre Empirical Methods of the Institute for Employment Research, and head of the Department for Product and Program Analysis of the German Federal Employment Agency.
Susanne has published books on survey sampling, statistical matching/data fusion, and prediction techniques, along with articles touching very different topics.
From 2007 to 2013 she was a member of the German census committee and served as a member of the German data council.
Susanne's particular research interests are
| Day | Topic | Details |
|---|---|---|
| 1 | Introduction to missing data, missing data mechanisms, patterns and overview of missing data techniques and applications in empirical settings. |
90min lecture/discussions + 90min. lecture/discussions. |
| 2 | Single imputation, maximum-likelihood methods for missing data, multiple imputation (MI) in general and applied multiple imputation |
90min lecture/discussions + 90min. lecture/discussions. |
| 3 | Basic MI algorithm concepts, proper MI, chained equations, Markov chain Monte Carlo, standard MI routines in SPSS, STATA and R (focus on R). |
90min lecture/discussions + 90min. lecture/discussions. |
| 4 | Algorithms for handling challenges in empirical settings, robust MI algorithms like predictive mean matching |
90min lecture/discussions + 90min. lecture/discussions. |
| 5 | Overview of diagnostics and evaluation methods, setting up a simulation study, relaxed repetition and summary |
90min lecture/discussions + 90min. lecture/discussions. |
| Day | Readings |
|---|---|
| 1 |
Little, R.J.A. & D.B. Rubin (2002). Statistical Analysis with Missing Data. 2nd edition, New York. – Chapter 1 |
| 2 |
Rässler, S., Rubin, D.B., Zell, E.R. (2013). Imputation. In: Wiley Interdisciplinary Reviews: Computational Statistics 5 (1), 20-29. Enders, C.E. (2010). Applied Missing Data Analysis. The Guilford Press. – Chapter 4. |
| 3 |
Little, R.J.A. & D.B. Rubin (2002). Statistical Analysis with Missing Data. 2nd edition, New York. – Chapter 10 |
| 4 |
Van Buuren, S. (2012). Flexible imputation of missing data. Chapman & Hall. – Chapter 3 |
| 5 |
Van Buuren, S. (2012). Flexible imputation of missing data. Chapman & Hall. – Chapter 5 |