Missing/Absent Data Treatment on JMP - Two-way ANOVA

[#1]

Hi there. I am a new user of JMP version 16.0. I am using student scores to analyze the prediction scores. I am utilizing the Fit-to-Model. There are data where students are absent during the test. Thus, I leave the score as a character such as 'ABS', set as numeric - continuous at the column setting.

I develop a linear model to analyze the prediction and the results are successfully be calculated by JMP. I wonder how the treatment is carried out for the absent/missing data? I tried to look for a JMP manual to understand the method, but I cannot find any. 

There are 2 cases where I'd like to understand the treatment of missing data. Kindly refer attachment for details.

Case 1: When calculating prediction for the current t (t=time) and the absent score is at t time.

Case 2: When calculating prediction for the next t (t=time) and the absent score is during previous t.

The linear model is;

Yijt = Grand mean + <student i> + <subject j> + <(subject*test)jt> + <(student*subject)ij> + εijt 

I attach the PDF to explain the simulation, and the operational data for JMP analysis with its results.

Thank you for your help and kind guidance. If there is any beneficial manual or equivalent to understand the treatment, please feel free to share it with me. It is highly appreciated. Thank you.

[#2]

You can see in the report that JMP is not using "the missing values", i.e. only 31 or 23 observations are used for the model.

Simply try to change the missing value to a non missing, and you will see the difference.

[#3]

BTW, there is a detailed description for Fit Model: Fit Model (jmp.com)

For modeling, JMP (and other software) does not need a fully occupied results matrix. It kind of "interpolates". You can see it also when switching extrapolation control on in Predction Profiler, then switching to the parameter setting with missing value.

[#4]

Hi Georg. Thank you for the feedback, it is highly appreciated. I tried to find the extrapolation control in the prediction profiler, but I cannot find any related menu. Where can I find the menu to switch on the extrapolation? Thank you for your kindness.

[#5]

Sorry, it is a JMP Pro function, there it is unter red triangle menu of Prediction Profiler.

In your case (small data set) it is not important to use it, because you know about missing data. If the data set is much larger, there are also other tools to graphically see, where data is missing, e.g. in Graph Builder.

[#6]

Dear Georg

Thank you so much for the information.

It is highly appreciated!