cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Try the Materials Informatics Toolkit, which is designed to easily handle SMILES data. This and other helpful add-ins are available in the JMP® Marketplace
Choose Language Hide Translation Bar
nickki_s
nickki_s
Level I

LSMeans Tests between Student's t vs. Tukey HSD

Dec 6, 2018 03:42 PM (7723 views)

Has anyone seen this result where Student's t letters are the same while Tukey HSD letters are different. I usually see visa versa. Any thought?

 

 

Preview file
230 KB
0 Kudos
1 ACCEPTED SOLUTION

Accepted Solutions
Mark_Bailey
Staff Mark_Bailey
Staff
Solution

Re: LSMeans Tests between Student's t vs. Tukey HSD

Dec 6, 2018 03:57 PM (7718 views)  |  Posted in reply to message from nickki_s 12-06-2018

I would not perform any post hoc tests until I validated the assumptions of the model and addressed all the issues with the estimation.

  • Did you notice the warnings in the report of the variance components?
  • Did you notice that your model is over-specified and the fixed and random effects are confounded?
  • Did you notice that one of the random effects is zeroed?
  • How do you justify treating gender as a random effect?

 

I don't think that this analysis is quite ready for any post hoc tests.

View solution in original post

1 REPLY 1
Mark_Bailey
Staff Mark_Bailey
Staff
Solution

Re: LSMeans Tests between Student's t vs. Tukey HSD

Dec 6, 2018 03:57 PM (7719 views)  |  Posted in reply to message from nickki_s 12-06-2018

I would not perform any post hoc tests until I validated the assumptions of the model and addressed all the issues with the estimation.

  • Did you notice the warnings in the report of the variance components?
  • Did you notice that your model is over-specified and the fixed and random effects are confounded?
  • Did you notice that one of the random effects is zeroed?
  • How do you justify treating gender as a random effect?

 

I don't think that this analysis is quite ready for any post hoc tests.