You are here: Home / The IPN / Archive / IPN Doctoral Student Awarded Faculty Prize by Kiel University

IPN Doctoral Student Awarded Faculty Prize by Kiel University

June 7th, 2018

Yesterday, Dr. Simon Grund was awarded the 2017 Kiel University Faculty Prize of Philosophy. The prize is annually awarded to an outstanding dissertation. It is endowed with 1000 Euro. The university president Prof. Dr. Lutz Kipp awarded the prizewinner during the ceremony yesterday evening.

Simon Grund, a member of the Department of Educational Measurement at the IPN, completed his doctoral thesis in the Department of Educational Measurement under Prof. Dr. Oliver Lüdtke (IPN) with the distinction "summa cum laude". The title of the thesis is: "Multiple Imputation of Missing Data in Multilevel Research".

In his dissertation he dealt with the treatment of missing values. In psychological research, empirical data often show missing values, for example when students skip a task in a test or do not answer certain questions in a questionnaire. Such an incomplete data set can pose major problems during evaluation. Statistical conclusions can be distorted if missing values occur systematically, so that, for example, students with poor performance are more affected by missing values. But even if missing values are completely random, they are a problem because they reduce the number of data points and thus the precision of the conclusions.

The statistical literature recommends various procedures for dealing with missing values. This includes the procedure of multiple imputation, with the help of which several plausible "substitutions" for all missing values are generated on the basis of (a) the observed values and (b) a statistical model (the imputation model). The multiple imputed data are then analyzed separately and the results are summarized to a final conclusion. Under certain conditions this procedure allows conclusions to remain undistorted despite missing values.

In addition, psychological data often have a hierarchical structure in which individuals appear "nested" in groups. Multilevel models are often used to analyze hierarchical data, allowing statistical coefficients to vary across groups. In the context of hierarchical data, multiple imputation is associated with special challenges, because the correct specification of the imputation model requires not only the data structure, but also the planned analysis. This is not always easy, especially with complex multi-level models. In his dissertation, he dealt in detail with the question of how missing values in hierarchical data can be handled reliably. In several papers he compared different approaches to the treatment of missing values using simulation studies and theoretical derivations and evaluated their properties in the context of various applications of multi-level analysis.

His dissertation also has a practical side: As part of his research work, he wrote the software package "mitml" for the statistical software R, which makes it possible to perform and analyze multiple imputations for hierarchical data and to summarize their results without requiring comprehensive statistical or programming knowledge.

This includes some of the above approaches to the imputation of missing values as well as the approaches to the use of multiparameter tests and model comparisons examined in a further paper. In this context, he has also written a tutorial that is intended to give applied scientists an introduction to the use of multiple imputation for hierarchical data.


Further Publications:

Grund, S., Lüdtke, O., & Robitzsch, A. (in press). Missing data in multilevel research. In S. E. Humphrey & J. M. LeBreton (Eds.), Handbook for multilevel theory, measurement, and analysis. Washington, DC: American Psychological Association.

Grund, S., Lüdtke, O., & Robitzsch, A. (2018a). Multiple imputation of missing data at level 2: A comparison of fully conditional and joint modeling in multilevel designs. Journal of Educational and Behavioral Statistics, 43, 316–353. doi:10.3102/1076998617738087

Grund, S., Lüdtke, O., & Robitzsch, A. (2018b). Multiple imputation of missing data for multilevel models: Simulations and recommendations. Organizational Research Methods, 21, 111–149. doi:10.1177/1094428117703686

Grund, S., Lüdtke, O., & Robitzsch, A. (2016a). Multiple imputation of multilevel missing data: An introduction to the R package pan. SAGE Open, 6(4), 1–17. doi:10.1177/2158244016668220

Grund, S., Lüdtke, O., & Robitzsch, A. (2016b). Pooling ANOVA results from multiply imputed datasets: A simulation study. Methodology, 12, 75–88. doi:10.1027/1614-2241/a000111