Sensorial Analysis - Duncan & Newman-Keuls tests

Question

Hey,

I have some potential customer in sensorial analysis who are interested by JMP Pro. they want to do some specific tests: Duncan test an Newman-Keuls Test. I don't find them in documentaiton neither in File exchange.

Does anyone has a knowledge of this tests and perhaps a workaround?

Thank a lot,

Florence

bernd_heinen · Answer

Hi Florence,

If you range the multiple test procedures with respect to their degree of conservatism you get this order:

Conservative: Duncan > Scheffé > Tukey > Newman-Keuls > Least significant difference > Not conservative.

JMP offers Tukey and Least significant difference (option “Each pair, Student’s t”), so Duncan will give less significant results than Tukey, whereas Newman Keuls is between Tukey and LSD. This may help to see what one could expect.

Theoretically all these tests use the same test statistic but only different parameters of its distribution to calculate the p values. In the JMP documentation there are literature links, so that one could dive into it and find the appropriate statistics. Programming should then be straightforward.

Best regards

Bernd

DonMcCormack · Answer

Florence,

They are both multiple comparison tests and are similar to Tukey's HSD or Dunnett's test found under Fit Y by X or Fit Model (when there are no random effects).

Don