topic Re: How to show the exact P-value when running Partial Correlation in Discussions

How to show the exact P-value when running Partial Correlation

nguyentandat234 — Tue, 07 Jan 2020 09:47:07 GMT

I just ran a Partial Correlation with 3 Variables : Paternal Care, Maternal Care and Self-Competence while controlling for Paternal Control and Maternal Control. I did this by going to Analyze-> Fit Model-> Add Paternal Care, Maternal Care and Self-Competence to Y while adding Paternal Control and Maternal Control to Model Effects.

I want to know the P-value of the Partial Correlation but can't find a way :(.

Re: How to show the exact P-value when running Partial Correlation

Mark_Bailey — Tue, 07 Jan 2020 12:38:26 GMT

I understand the model that you specified but I do not understand what you mean by 'partial correlation' in this context.

Re: How to show the exact P-value when running Partial Correlation

statman — Tue, 07 Jan 2020 18:52:44 GMT

I'm with Mark on this....but why do you need exact p-values? You realize these are estimates at best based on the data and comparisons you are making. You should read: The American Statistician journal titled "Statistical Inference in the 21st Century: A World Beyond p < 0.05". Here is the link:

https://www.tandfonline.com/toc/utas20/73/sup1?nav=tocList

Re: How to show the exact P-value when running Partial Correlation

nguyentandat234 — Wed, 08 Jan 2020 03:41:25 GMT

Hi, sorry for the confusion. What I meant is the P-value of the Partial correlation you get when you try to run MANOVA. It doesn't need to be the exact P-value. I just want to know whether the correlation was significant or not. As you see in the image, the result only show the correlation without giving any indication about the significance of the result. :(

Re: How to show the exact P-value when running Partial Correlation

Mark_Bailey — Wed, 08 Jan 2020 10:59:19 GMT

I guess I did not understand your model after all. You did not mention that you were using MANOVA.

The hypothesis tests for the partial correlations are not available at this time. You might contact JMP Technical Support (support@jmp.com) to see if there is a work-around available. That is, you might be able to determine the p-value from the information given in the MANOVA report.