What is it about?
Many experiments involve assessing the effects of several inputs (factors) on one or more outputs (responses). Good design involves choosing to run combinations of factor levels which will make the data analysis as informative as possible. Much recent work has developed designs for situations in which some of the factors have levels which are harder to set than others, so they are varied less often. Most of this has emphasised estimation. This paper extends this work to consider data analysis for carrying out inferences such as hypothesis tests and confidence intervals.
Featured Image
Why is it important?
This paper allows designs to be better tailored to the data analysis which will actually be performed.
Perspectives
The work presented here is a big step to making optimal designs more like classical designs in the sense that they have many good properties. It also encourages experimenters to think carefully about the data analysis they will perform before they design the experiment.
Steven Gilmour
King's College London
Read the Original
This page is a summary of: Split-Plot and Multi-Stratum Designs for Statistical Inference, Technometrics, April 2017, Taylor & Francis,
DOI: 10.1080/00401706.2017.1316315.
You can read the full text:
Contributors
The following have contributed to this page
