cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
JMP is taking Discovery online, April 16 and 18. Register today and join us for interactive sessions featuring popular presentation topics, networking, and discussions with the experts.
Choose Language Hide Translation Bar
Faizan
Level I

Linear mixed effects models

I am a rooky in statistics and this is the first time I'm using jmp, but I have some knowlegde about STATA. I first tried to do it by myself, but it's too hard. I hope some of you could help me with the following:

 

I am trying to set a Linear mixed effects models. I have data at 3 different time points (n=16). My response variable (Y) is continous at the last time-point and my predictors include 2 categorical variables [A, B] (each have three different levels).

 

What I want to do is, is analyzing Y  over time by A B  categories with random effects for subject. Within A and B i want to compare differnt subcategories with the reference one. e.g. A0 vs. A1 and A0 vs A3. How i can specify the reference category for my categorical variables ?

 

Then I want to report  differences and 95% confidence intervals.

1 ACCEPTED SOLUTION

Accepted Solutions
Georg
Level VII

Re: Linear mixed effects models

@Faizan when starting with JMP, it May also be an Option to look at data first (Gaphbuilder ..) and then start Y by X e.g. with local data filter … or heat map to understand what the data contains before running the full modeling machine. I can not really imagine how your full data set Looks like, you sent only a small subset I think. JMP offers plenty Methods to understand data.

As @Mark_Bailey stated, the fit model platform will be your way to perform the Task. I put some scripts into your data table that you can see how it May look like.

 

What do you mean by reference? Is this a certain Setting (=Level) of your categorical variable A+B?

In the fit x by y platform you can see the answer Y depending on the Settings of A or B, Controlling with the Local data filter what part of data you are Looking at. You can mark the rows containing the reference Settings, and Show it up in the Analysis, is that what you are Looking for?

Georg

View solution in original post

3 REPLIES 3

Re: Linear mixed effects models

Exactly what is your model? What would you specify in STATA for your purpose?

 

You can start with Analyze > Fit Model. See Help > JMP Documentation Library > Fitting Linear Models. You can define the linear predictor with factors, cross terms, nested effects, and random effects here.

 

You would want to fit the model and then use custom tests (i.e., contrasts) to test the null about the reference category.

Georg
Level VII

Re: Linear mixed effects models

@Faizan when starting with JMP, it May also be an Option to look at data first (Gaphbuilder ..) and then start Y by X e.g. with local data filter … or heat map to understand what the data contains before running the full modeling machine. I can not really imagine how your full data set Looks like, you sent only a small subset I think. JMP offers plenty Methods to understand data.

As @Mark_Bailey stated, the fit model platform will be your way to perform the Task. I put some scripts into your data table that you can see how it May look like.

 

What do you mean by reference? Is this a certain Setting (=Level) of your categorical variable A+B?

In the fit x by y platform you can see the answer Y depending on the Settings of A or B, Controlling with the Local data filter what part of data you are Looking at. You can mark the rows containing the reference Settings, and Show it up in the Analysis, is that what you are Looking for?

Georg
Faizan
Level I

Re: Linear mixed effects models

@Georg @Mark_Bailey  thank you for suggestion. The scripts helped me to understand.