nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Analyze Latin squares using JMP

paul1

Hi,

I would like to analyse some data that is based on a Latin squares design. I have imported the data into JMP, two columns corresponding to the separate blocking variables, one column corresponding to the treatment variable and one column corresponding to the response. However, I am unable to specify more than I blocking column in the Fit Y by X platform. I searched through the discussion archives but could not find an answer, and would greatly appreciate any help.

Best,

Paul

Accepted answers

Mark_Bailey

Jim is correct. The Oneway platform (through Fit Y by X) is intended for a one-way analysis of variance with the exception of including a single blocking variable. This requires a special role in Oneway. The more general Fit Least Squares platform (through Fit Model) is what you want.

Blocking is not a 'role' here as it is in Oneway. Instead, you simply enter the term for the treatment factors and each blocking factor, two of them in the case of your Latin square design.

The only other distinction is if you see the effect of the blocking as fixed or random. So far, JMP will model it as a fixed effect. If you want to estimate it as a random effect, then select the term in the list of effects and the click the red triangle for Attributes and select Random Effect.

All comments

txnelson

You will want to use the Fit Model Platform.

paul1

I can add the response column to Y, and the treatments column to 'Construct Model Effects', but I am still not clear on how to set the other two columns as blocking.

Best,

Paul

Mark_Bailey

Blocking is not a 'role' here as it is in Oneway. Instead, you simply enter the term for the treatment factors and each blocking factor, two of them in the case of your Latin square design.

bordini

Please, would anyone be able to tell me the steps to run a latin square and greak latin square in JMP? I understand that I need to use the fit model function, but in which fields should I add the treatments and the block factors? Are all of them added together in "Construct Model Effects" or somehow doI need to point out which one is my treatment?

I'm sorry if my question is silly. I'm very new in statistics and especially in JMP.

Thank you

Mark_Bailey

There is nothing special about a Latin square design as far as analysis in JMP. Assuming that you already have your data table arranged with each run on a separate row and each factor and response in a separate data column, you can follow this procedure:

Select Analyze > Fit Model.
Select the data column with the response measurements and click Y.
Select the data columns with the factor levels and click Add.
You can include terms for interaction and non'linear effects using Cross or the convenient Macros, but you didn't indicate a need for those terms.
You can control the initial report by changing the Emphasis.
Click Run.

See also, Help > Books > Design of Experiments for many examples of analyzing a DOE.

bordini

Thank you very much for your reply! I appreciate your help and time.

Kieu

This might be too late for you but I hope it helps someone since it took me a while to figure out.

To run Latin Square in JMP, you must have Row, Column, Treatment and the Y-response.

You must set your data type correctly for all of the variables (Row, Column and Treatment). You must change the column information, double click on the column: Data type as Character Modeling type as Nominal. If your x variables are in numbers/continuous it will run as regression.

Then Analyze/ Fit model/ put Y in Y and all the Row, Column and Treatment under "Construct Model Effects", keep all others as default. Run

Click on Response Y red triangle/ Estimates/Multiple comparisons to get the LSD for your treatment. If you use student's t, then check it. Click on red triangle of Student's t to get All pairwise differences connecting letters.

Kieu