Students T post HOC test

Alec1293 — Sun, 29 Apr 2018 15:58:42 GMT

I am running a mixed model in JMP, there is a three way interaction I am interested in the outcome of, which is the interaction: Location * Date * Treatment. I am looking for differences between locations and treatments on each date but unfortunately I cannot use Test Slices because there are 6 comparison within each day I am interested in. So what I have to do is use the LSmeans Students T which gives every possible comparison (nearly 400) but I am only interested in what happens within each day (6 each day, 7 days = 42 comparisons) and maybe what happens from one day to the next but not the first day vs. last day.

I am using a bonferroni correction to correct for the multiple comparisons by doing alpha(0.05)/ k (# of comparisons being considered), this provides a corrected p-value for the multiple comparisons. Which I know protects agains type I error but adds vulnerability to type II error possibility. Is this the best way to go about this?

Thank you for any help.

Re: Students T post HOC test

Alec1293 — Sun, 29 Apr 2018 21:49:03 GMT

I believe this correction may be to conservative though, in one case I am correcting for 126 comparisons. Is there a better correction method I can use for multiple comparisons?

topic Re: Students T post HOC test in Discussions

Students T post HOC test

Re: Students T post HOC test