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
Kynda
Level III

How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analysis?

I've performed Repeated Measures Analysis (Mixed Model) for my data [I've 3 groups (Control, 1mM treatment, 2mM treatment), Each group has two replications (R1 & R2) and I take measurements at 4 time points (0d,1d, 3d,5d). Each replication has six samples (subsampling) and same sample use for all the time points (repeated measures)]. (file attached)

--------------------------------------------------------------------------------------------------

This is how I perform the analysis:

Analyze > Fir Model 

EG1 to Y

Group, Rep, Time added to Construct model

Then, Rep(Group) & Random (by changing attributes to Random effect) 

Method: REML

Run!

--------------------------------------------------------------------------------------------------

Results output looks like:

1..png2..png

 

I want to draw a graph like below and denote significance difference at each tim epoint? 

Picture1.png

 

I want to get the mean separation for each time point separately, and denote in the graph?

Time"0": what is the mean separation of Control /1mm/2mM

Time "1": etc.

And please tell me how did you create this graph in JMP and meanseparation for eachtime point separately?

 

Thank you for your kind support

 

 

2 ACCEPTED SOLUTIONS

Accepted Solutions

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

Hi Kynda,

The graph I created in Graph Builder is really just a couple of drags for you data.  Your data is set up nicely in a stack.  Drag EG1 to the Y axis, Drag Time to the X axis.  Drag Group to Color by on the right and then drag Group to Overlay on the right.

 

On the left side of the set control window under Bar is where you can change the view by bar style, summary and error bars.  For the graph shown below I have chosen Side by side, Auto, Mean, Standard Deviation, and Label by Variable.  

Capture1.JPG

 

 

 

 

 

 

 

 

 

 

 

Hope this helps.

Bill

View solution in original post

Kevin_Anderson
Level VI

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

Hi, Kynda!

 

I took the dataset you attached and added some scripts to it.

 

The first one analyzes the data as a replicated repeated measures experiment.  Warning: I know next-to-nothing about the design you ran to gather these data, about how the data were gathered, or basically anything about your experiment.  The model doesn't converge well, and throws a warning about it.

 

Under the Little Inverted Red Triangle (LIRT) next to the Fit Mixed header in the Analysis window, choose Model Dialog, and you can review how I set up the model.

 

Look down in the analysis, and find the output on the Multiple Comparison:

image.png

 

 

 

 

Note the Quantile and the Adjusted DF.

 

In a new datatable, enter this formula:

image.png

 

 

 

...this is how I got the weird confidence interval number (0.9888777717).

 

The analysis window gives the Pairwise Differences...I think this is what you're calling the "mean separation".

 

Run the second script, and you get the graph:

image.png

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

If you right-click on the scripts and edit them, you can review the platforms and set-up for performing the analysis and the graph with a point-and-click interface.

 

Good luck!

View solution in original post

8 REPLIES 8

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

Hello,

I created the graph below in Graph Builder.  The colors can be changed as needed or wanted.

Capture.JPG

 

 

Kynda
Level III

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

Thank you @Bill_Worley Appreciated. 

I want to get the mean separation (Tukey test) for each time point separately by JMP, and denote in the graph? 

Time"0": what is the mean separation of Control /1mm/2mM

Time "1": etc.

And please tell me how did you create this graph in JMP and meanseparation for eachtime point separately?

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

Hi Kynda,

The graph I created in Graph Builder is really just a couple of drags for you data.  Your data is set up nicely in a stack.  Drag EG1 to the Y axis, Drag Time to the X axis.  Drag Group to Color by on the right and then drag Group to Overlay on the right.

 

On the left side of the set control window under Bar is where you can change the view by bar style, summary and error bars.  For the graph shown below I have chosen Side by side, Auto, Mean, Standard Deviation, and Label by Variable.  

Capture1.JPG

 

 

 

 

 

 

 

 

 

 

 

Hope this helps.

Bill

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

I forgot to add that you will need to select the Bars icon at the top of the window once you have your data in place. The icon you need to choose is highlighted in blue at the top of the image.
Kynda
Level III

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

Thank you very much @Bill_Worley those steps help a lot.

I am sorry, if my English makes you confuse.

 

My interest is to perform the Tukey test (multiple comparision /mean separation)  for each time point separately by JMP and denote in the graph. 

As shown in these below bar charts, I want to show significant difference between each bar at each time points.

 

 

Image source

 

 For this I have to get connecting letters report for each time point (e.g. below image)

 

Image result for mean comparison all pairs tukey jmp letters

Image source

So my exact question is, after perform Repeated Measures Analysis (Mixed Model) ANOVA, (assume my data is normal distributed for this question), How can I perfrom multiple comparision in JMP for different time points?

Is this possible in JMP?

When I do this I get mean separation as whole, not for different time points.

 

Could you please help me witht his.

 

Thank you.

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

Hi Kynda,

 

I cannot come up with a quick and easy way to do what you are asking.  You can run your ANOVA and then right click on the connected letters report and either click Make into a Data Table or control click and select Make a Combined Data Table.  You then have to wrangle the data little more by combining the different letter columns, but it still doesn't give you or get you everything you want or need to see.

 

Another option to try is use Fit Model, build the model to predict EG1 and then click the red hot spot by Response at the top.  From there go to Estimates > Multiple Comparisons.  From this option you can choose which factor you want to make the Tukey pairwise comparisons for, but you will not get a connecting letters report.

 

Maybe someone out there has a better solution for you.

 

Best,

Bill

Kevin_Anderson
Level VI

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

Hi, Kynda!

 

I took the dataset you attached and added some scripts to it.

 

The first one analyzes the data as a replicated repeated measures experiment.  Warning: I know next-to-nothing about the design you ran to gather these data, about how the data were gathered, or basically anything about your experiment.  The model doesn't converge well, and throws a warning about it.

 

Under the Little Inverted Red Triangle (LIRT) next to the Fit Mixed header in the Analysis window, choose Model Dialog, and you can review how I set up the model.

 

Look down in the analysis, and find the output on the Multiple Comparison:

image.png

 

 

 

 

Note the Quantile and the Adjusted DF.

 

In a new datatable, enter this formula:

image.png

 

 

 

...this is how I got the weird confidence interval number (0.9888777717).

 

The analysis window gives the Pairwise Differences...I think this is what you're calling the "mean separation".

 

Run the second script, and you get the graph:

image.png

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

If you right-click on the scripts and edit them, you can review the platforms and set-up for performing the analysis and the graph with a point-and-click interface.

 

Good luck!

Kynda
Level III

Re: How to perform mean separation (Tukey's procedure) for sub categories / Repeated Measures Analys

Thank you @Kevin_Anderson and @Bill_Worley taking your time to help me.

 

@Kevin_Anderson

As you've noted,

Picture1.png

 

I can not find this in the file you've attached. My file looks like below image;

Picture1111.png

I got told from statistician, as I am taking 6 samples at each time point I have to include Rep*Time(Group)&Random then.  That’s really my experimental error term. To do the analysis that I want, I have to include the interaction between Group and Time.  Doing Tukey comparisons in this way will be overwhelming since it will consider all 12 choose 2 combinations of group and time.  The only way around this in JMP is to do some scripting myself.

Unfortunately I am not expert in this programming. Would be greatful if you could help me with this steps. 
Since most of my experiments are looking like this, this will be invaluble to perfrom the analysis by JMP.
 
Thank you!