Tarone-Ware Two-Sample Test
Tarone-Ware Two-Sample Test
See Also:
Description:
The Tarone-Ware Two-Sample test is a non-parametric test for comparing an individual compliance well to pooled background well data.
Use:
Use the Tarone-Ware test to compare individual compliance well data to pooled background data. The test is suitable for non-normally distributed data or data sets with a high percentage of non-detects. The test is a suitable alternative to the Wilcoxon rank-sum test. However, unlike the Wilcoxon test, this test does not use ranks and may therefore produce different outcome on log-transformed data or when non-detects are replaced with 1/2 of the detection limit.
Remarks:
This test can be run at the 95% or 99% confidence levels. The Unified Guidance presents an example where the test is run at the 95% confidence level. However, 95% will likely results in an excessively high false positive rate. As an individual well comparison, 99% confidence is more appropriate.