How to Derive the Correlation Between Two Continuous Variables with Corrections for Covariates (JMP 16.1, Window 10)?

[#1]

Hi JMP Community,

This question might be nonsense, so bear with me.

I need to estimate the degree of association between a laboratory measure and a clinical endpoint while considering the effect of age and gender (see attached).

1. I know that the laboratory measure correlates significantly with the clinical endpoint without any correction (Pearson's method)
2. If I create a simple linear model with the Laboratory Measure, Age, Gender, and their interactions, I confirm the effect of the Laboratory Measure on the Clinical Endpoint and the effect of Age.
3. However, I have not found an option in the Fit Model platform that will return the corrected correlation ratio for each variable.

Is this a legitimate question, and if it is, is there an option to conduct this type of analysis in JMP 16.1 (not Pro)?

Thank you.

Best,

TS

[#2 · Accepted solution]

HI @Thierry_S ,

You might try fitting a model of your endpoint vs the covariates (Age, Gender), then saving the residuals. Then look for the correlation between this "corrected" endpoint and the lab measure using Fit Y by X.

I have attached a version of the table with saved scripts for this analysis.

To be clear: I am not advising that this is the appropriate analysis or that it would be compliant with regulations for your industry.

I hope this helps,

Phil

[#3]

Hi @Phil_Kay 

Thank you for the workaround. This is the approach I imagined I would need to take. In future versions of JMP, it may be possible to conduct this type of analysis within the Fit Model platform.

Thanks.

Best,

TS

[#4]

I suggest that you add this to the JMP Wish List, @Thierry_S .