Semi-partial correlations

Discuss statistics related things

by keithions » Wed Apr 07, 2021 10:06 am

Using jamovi to do a semi-partial correlation produces two values for r and two associated p values.
I assume this because the programme does not know which is the predictor variable from which to ‘partial out’ the control variables from.
However, I don’t know which of the r values is the one of interest in the table provided by jamovi, and I wondered if someone would tell me how to decide which is the correct value?
keithions
 
Posts: 2
Joined: Wed Apr 07, 2021 10:02 am

by Ravi » Wed Apr 07, 2021 11:03 am

Hi Keith,

So the table has a footnote that describes for which variable the variance of the control variable is partialled out:
Screenshot 2021-04-07 at 13.00.21.png
Screenshot 2021-04-07 at 13.00.21.png (116.16 KiB) Viewed 582 times


So the variance of the control variables is partialled out for the "column" variables, but not for the "row" variables in the semi-partial correlation table.

Does that make sense?

Cheers,
Ravi
User avatar
Ravi
 
Posts: 163
Joined: Sat Jan 28, 2017 11:18 am

by keithions » Wed Apr 07, 2021 2:03 pm

Yes thanks, Ravi. That is very helpful.
Thanks
Keith
keithions
 
Posts: 2
Joined: Wed Apr 07, 2021 10:02 am


Return to Statistics