is Partial Correlation working?

[coledavis]

Hi. I'm updating my statistics books, but datasets which work in R and in JASP don't seem to elicit any response from Jamovi's Partial Correlation routine. www.vorpress.com

[jonathon]

hi cole,

it looks to be working here. could you attach a .omv file here with the analysis in the "not working" state?

with thanks

[coledavis]

Hi Jonathon - try this. I've included a correlation matrix as well.

[jonathon]

perhaps i don't understand partial correlation ... but i would expect that you would need to provide multiple variables in the 'Variables' box. the attached file only contains 'Violence' in the 'Variables'.

let me ask ravi about this.

jonathon

[Ravi]

Yeah, the partial correlation calculates the correlation between the variables in the "Variables" box, with the variance from the variables in the "Control Variables" (semi) partialled out. So you'd always have to add at least two variables to the "Variables" box. I'm curious to know what R and JASP do differently here?

[coledavis]

Thank you, my mistake, as variables wrong way round. The reason this happened was that I was assuming that it would be like JASP's interface, where partials are part of the linear regression tool. I attach a picture of the interface.
From what I can see, the results are different between JASP and jamovi. Jamovi and R work out ok, so I presume that the difference from JASP is that it uses linear regression, not conventional partials. I'll send a file showing these.

[coledavis]

Files sent to you direct.

[jonathon]

thanks cole,

so is this all sorted then?

jonathon

[coledavis]

Sorry, didn't see your last message. Yes, thanks.