How do I compare differences between LS-means

[#1]

Is there a way to compare the differences between the LS-means of several groups? In other words, how can I detect differences between the differences in different groups? If I run a linear model on my original data, I can use multiple comparisons to get the difference between the LS-mean of each group and the standard errors of those differences. Can I go a step further and do another test to see, for example, if the difference in the LS-means of groups A and B is different than the difference in the LS-means of group C and D?

[#2]

Hi @jeekhoff 

my approach to this would be to use the groups as an ordinal variable and order the categories by their LS means. Once you have that in a model, you can use custom tests to compare all sorts combination of the coefficients (differences between categories)

let us know if it works

[#3]

Yes, you want to sue a contrast. See this section of JMP Help for Fit Least Squares that you are using to get the LS Means.