Essays on Inference in Linear Mixed Models
by Peter Kramlinger
Date of Examination:2020-04-28
Date of issue:2020-05-11
Advisor:Prof. Dr. Tatyana Krivobokova
Referee:Prof. Dr. Tatyana Krivobokova
Referee:Prof. Dr. Stefan Sperlich
Files in this item
Name:Dissertation_Kramlinger.pdf
Size:646.Kb
Format:PDF
Abstract
English
This work addresses two aspects of inference in linear mixed models. It is based on two manuscripts, which are provided as addenda. The first manuscript deals with the distinction between marginal and conditional multiple inference, and provides confidence sets useful for testing in each scenario. The second manuscript is concerned with the issue of constructing confidence sets based on the lasso, which hold uniformly over the space of coefficient and covariance parameters.
Keywords: Small Area Estimation; Marginal Inference; Conditional Inference; Multiple testing; simultaneous inference; Lasso; Sparsity; REML; Variance components