AOAC Gluten Qualitative Validation Guidance-FINAL (July 2023)

Gluten Working Group Comments/Responses (Round 1)

o Need to check with Paul on the Box, Hunter & Hunter reference. Perhaps the general design needs to be explained in another appendix. The layout I have seen used has this type of data table (except there would be ten replicates instead of the 5 used here):

A logistic regression is then run with POD as the dependent variable (see next page), and each of the parameters as factors (again, sample sizes would be ten instead of the five shown here). You can see in the ANOVA if there were any significant differences, and the sum of squares data can show which factors contributed to that – in this example, strip incubation time had a large effect, and sample volume changes were also significant.