| Search website: | Search project:
Viewing 42 out of 42 projects
| ||||||
|
|
| ||||||
funding agency
Multiway nonparametric Scheirer-Ray-Hare ANOVA
| Acronym: | NonParametric |
| Project type: | Consulting RTD |
| Time frame: | 2003 |
| Funding agency: | UGent |
| Geographic keyword: Europe | Belgium | |
| General keyword: Software development | |
| Specific keyword: Large multidimensional data sets | Scheirer-Ray-Hare test |
Scheirer-Ray-Hare Project: Non-parameteric Multiple-way Analysis of Variance - testing two or more independent factors at once
The software developed within the framework of this project encompasses following functionality, where parametric methods start from the boundary condition that measurements are supposed to be Gaussian distributed and non-parametric methods are suitable for non-gaussian type of random variables:
Multiple-way ANOVA - No replication - Parametric:
This design is used when there is (a) not enough time or resources for replication for each factor (2 or more factors) level combination, or (b) if the measurements are so similar that there is effectively no variation within each "cell" in the design, or (c) as the extension to the paired t-test for more than two observations (repeated measures design).
Multiple-way ANOVA - No replication - Non parametric:
Design as above. This design can be analysed with a Friedman test. The data must be set up in a row/column format with each row representing one level of one of the factors (e.g. strain or temperature). The test will then rank the data within columns and the probability that the columns are different is given.
Multiple-way ANOVA - Replication - Parametric:
There are two or more independent factors expressed as two or more levels. This time it is possible to get a probability for the interaction term as well as for the factors on their own (so three null-hypotheses to test in the Two-way case).
Multiple-way ANOVA - Replication - Non parametric:
Design as above. Original measurement values are assigned ranks in an extension of the Kruskal-Wallis test called the Scheirer-Ray-Hare test.
Legend to the figures:
Figure 1:
Figure 2: