Abstract
A method is developed and studied for testing equality of variances based on simplicial data depth and Mood's nonparametric test in the case of two samples. A method for calculating univariate simplicial data depth using a rank transformation is introduced. Type I error rates and power curves are compared for three existing tests for equality of variances and the data depth test using data simulated from the nonnal distribution and 5 nonnormal distributions. In addition, a new method of aligning two samples with unequal location parameters is proposed. This method shows significant improvement over aligning by either the median or mean in controlling Type I error rates of skewed distributions.
Keywords
Equal Variance Tests, Nonparametric, Mood's test, Simplicial Data Depth, Two Sample, Univariate Response
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Recommended Citation
McGaughey, Karen J. and Milliken, George A.
(2003).
"VARIANCE TESTING WITH SIMPLICIAL DATA DEPTH,"
Conference on Applied Statistics in Agriculture.
https://doi.org/10.4148/2475-7772.1182
VARIANCE TESTING WITH SIMPLICIAL DATA DEPTH
A method is developed and studied for testing equality of variances based on simplicial data depth and Mood's nonparametric test in the case of two samples. A method for calculating univariate simplicial data depth using a rank transformation is introduced. Type I error rates and power curves are compared for three existing tests for equality of variances and the data depth test using data simulated from the nonnal distribution and 5 nonnormal distributions. In addition, a new method of aligning two samples with unequal location parameters is proposed. This method shows significant improvement over aligning by either the median or mean in controlling Type I error rates of skewed distributions.
