Mathematics
8 years ago

Randomization tests recommended for evaluating group variance equality.

Journal of Statistical Computation and Simulation
Testing multiple-group variance equality with randomization procedures
Harry Charway, A. John Bailer

Paper Summary

Related papers

Theme background image
Mathematics
5 years ago

Preliminary tests for population variances unreliable under non-normality, study finds.

Skew Distributions
Sensitive Surveys
Missing Data
Theme background image
Agricultural and Biological Sciences
7 years ago

F test unreliable for comparing variances, leading to inaccurate results

Pollinator Decline
Population Genetics
Genetic Adaptation
Theme background image
Mathematics
8 years ago

New nonparametric Levene test improves accuracy of statistical analysis

Clinical Trials
Multicollinearity

The article tested different statistical tests for comparing variances in multiple groups. Randomization tests were found to be the most reliable, maintaining correct error rates in various conditions. Levene's test was too cautious, while Bartlett's and Hartley's tests were too lenient depending on the data distribution. Bartlett's test worked best with normal data. Overall, using randomization tests is recommended for accurately testing variance equality.