Collinearity statistics in JASP regression module?

PerPalmgren

Hi,

One question! Why are the collinearity statistics (VIF and Tolerance) in the regression module the same (see attached file)? How are these calculated? These statistics are not the same when I try to run the same data in other programs like SPSS, DATA and Jamovi.

Picture1.zip

All the best

Per (the JASP lover)🤩

EJ

I forwarded this to the team

EJ

PerPalmgren

@EJ

Hi,

Do you know if the JASP team has looked into this? I have had some students asking why the Collinearity statistics (VIF & Tolerance) in JASP gets different from when running the same procedure in other programs such SPSS, STATA, Jamovi, DATAtab. I cant answer why!🙄

Kind regards

Per

PerPalmgren

vandenman

Hi PerPalmgren,

Thanks for pointing this out.

It looks like we used a different method to compute the VIF than the other programs. The results were the same when there are only continuous variables, but as soon as there are categorical predictors (i.e. factors) then the results of JASP differed from the other programs. This will be fixed in the next release.

Cheers,

Don

PerPalmgren

Thanks Don!

Question. Is this rectified in JASP 0.17.2?

All the best

Per

vandenman

Hi PerPalmgren,

Unfortunately, it did not make it. It is now in the development version of jasp (so also in the nightlies) and it should also be in the next release.

Best,

Don