GraphPad Statistics Guide

False Discovery Rate approach to multiple comparisons

False Discovery Rate approach to multiple comparisons

Previous topic Next topic No expanding text in this topic  

False Discovery Rate approach to multiple comparisons

Previous topic Next topic JavaScript is required for expanding text JavaScript is required for the print function Mail us feedback on this topic!  

When you choose the False Discovery Rate (FDR) approach to multiple comparisons after ANOVA, Prism does the following:

1.Perform the comparisons you requested using the unprotected Fisher's LSD test. This results in a P value for each comparison. These P values do not correct for multiple comparisons. They are not multiplicity adjusted P values.

2.Use the FDR approach you chose (Prism offers three variants) to decide which P values are small enough to be deemed "discoveries". This calculation depends on which method you chose, and the value of Q you chose (the desired false discovery rate, as a percentage).

3.For each comparison, also compute a q value. The q values will be different for each comparison. If you had set Q to this value (what Prism reported as q) then this comparison would have been right on the border of being a "discovery" or not.

Notes:

When you choose the FDR approach, Prism will not report anything about statistical significance, and will not (cannot) report confidence intervals or multiplicity adjusted P values. But it does report q values, which are similar to adjusted P values.

While the FDR approach is often used to deal with many P values such as those computed by Prism's multiple t test analysis, they are not commonly used as followup tests for ANOVA.

The variable q is used as part of the results of the FDR approach to multiple comparisons, and as part of the Tukey and Dunnett multiple comparisons tests. The three tests define the variable q differently so they should not be compared.