How is JMP STD error for means and STD error for Mean difference calculated? I wish validate the numbers I get in Excel with JMP

[#1]

How does JMP calculate the Std Error of means and Std Error of Diff in the LS means Table and LS Means Differences Student's t.

I tried to replicate the results by STD error calculated from both Standard deviation of sample and population,

but the values do not match.

Please refer the document attached, I calculated the  STD errors from the values in 4 groups as well as the

STD error of the 4 means in in my validation data set. I am not able to match it with the output of JMP.

Please point me to where I am going wrong.

Thanks

[#2]

Do you mean to say that you started with the sample or population standard deviation of the response? That approach is invalid. That approach will lead to an inflated standard error. Why? Because that measure includes the fixed effects contributed to the variation in the response.

The mean sum of squares error or the root mean sum of squares (error) is the estimated variance or standard deviation, respectively, of the response without the fixed effects.

[#3]

I understand your concern, I am trying to replicate the workflow of my colleague and automate it. The first step for me is to understand how JMP gets the std error of the LS means as well as the stderr of diff of means at various levels in the studentized t-test. I understand this approach might not be valid, but for me to bring that up, I need to understand their process and replicate it first 

[#4]

Also please help me understand what formula I should use to measure the contribution of the fixed effects to the variation in the response,
So that I can replicate the JMP results.

[#5]

Hi,
In the Analyze > Fit Model platform, Standard Deviations are pooled across levels, hence Standard Error of individual levels will not match exactly.
If you are interested in calculating the Std Dev and Std Err for each level, I would recommend using the Summarize platform
Best,
TS

[#6]

Hi Thierry_S,

Thank you very much for your reply. I am actually interested in repeating the results in the document in excel. I am are trying to emulate the JMP Fit Model and the Studentized T-test results my colleague uses and unfortunately I can not use a different method.  I will have to understand why these results.

If the std deviations are pooled across the levels, why do the values not match the values I get in Excel? 

Thanks for helping me understand this.