Knowledge Discovery Group

An R-toolbox for Non-parametric Analysis of Split-Plot Designs

Description

Split-plot designs are an important method in experimental research. A split-plot experiment allows to investigate the effects and interaction of two independent variables while at the same time being not too expensive w.r.t. the number of participants needed. For analysing such experiment designs, split-plot ANOVA can be used if the data fulfills certain criteria such as normal distribution. When these criteria are not fulfilled, a non-parametric analysis of split-plot designs needs to be applied. However, existing statistical tools do not provide a good support for analysing such experimental setups.

Goal of this thesis is to develop an extension to the open source toolbox R for statistical computing (http://www.r-project.org/) to provide explicit support for split-plot experiments. Based on the literature, a R package will be designed, implemented and tested. In addition, we will compare the non-parametric analysis of split-plot designs against split-plot ANOVA.

In more detail, the work should cover:
- Learning about experiment designs and statistical analyses
- Implementing a non-parametric analysis of split-plot designs
- Comparison to split-plot ANOVA
- Programming in R

Requirements

- Knowledge about statistical methods is an advantage
- Interest in experimental design and statistical analyses

Got interested? Send an email!


News

  • Homepage kicked off!