Dixon's Test for Outliers

For 3 to 25 Samples

 

See Also:

RosnerTest for Outliers

Rejecting Outliers

Outlier List Report

Discordance Outlier Test

 

Description:

 

Dixontest provides a method of screening for outlier concentrations for data sets with 25 or fewer measurements. The method is iterative. In each iteration of the test, the highest or lowest outlier value is revealed. The next iteration is performed on the remaining values. Iterations continue until no data are shown to be outliers.

 

In each iteration, the highest and lowest critical values are calculated using a formula selected based on the number of data not yet shown to be outliers. These formulas are provided by Gibbons (1994). The critical value is then compared to tabulated comparison values based on the number of measurements now yet shown to be outliers, and the level of significance.

 

In ChemStatimplementation, Dixontest can be performed on all wells, all compliance wells, all background wells, or the selected well. This option is available from the right-click menu accessed over the Dixontest window. Remember that the total number of measurements screened can not exceed 25. Use Rosner's test for greater than 25 measurements.

 

ChemStat performs Dixontest at either the 1% or 5% levels of significance. This option is selected from the right-click menu accessed over the Dixontest window.

 

 

Use:

 

As a method of screening for outlier concentrations for data sets with 25 or fewer measurements.