[R] CORDIF test

[Marc Schwartz]

On Apr 2, 2014, at 8:09 AM, Elizabeth Caron-Gamache <babeth_013 at icloud.com> wrote:

> Hi, 
> 
> I search on your website for a definition of the CORDIF test, but it wasn’t successful. I’m analyzing an article that use that test and it’s not really documented on the net. The article refers to your website, so I pretend that you will be able to give me a brief explanation of this test. Here is the cote that talk about this test in my article : 
> 
> ‘' To compare these regressions and to see which—either body height or LLL—is best related to performance (Pearson correlation coefficients comparison), a CORDIF test (R software [www.r-project.org], multilevel package, ver- sion 2.12.1) was performed.
> 
> Does it use parametric or non-parametric values ?
> Is it a test to compare 2 groups only or it can be used for a comparison of more than two groups ?
> Why is it so hard to find information on that test on the net ?
> 
> Thanks for your time
> Have a nice day 
> 
> Elizabeth Caron
> Physical therapist student, Laval University, Qc, Canada


Thanks for including the citation, which indicates that the CORDIF test is part of the 'multilevel' package, which is on CRAN:

  http://cran.r-project.org/web/packages/multilevel/index.html

The reason that it is likely difficult is that 'cordif' is an abbreviation for "correlation difference", not the proper name for a test.

If you review the provided documentation for the package:

  http://cran.r-project.org/web/packages/multilevel/multilevel.pdf

you will see that there is a description of the cordif() function and a reference given:

Cohen, J. & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd Ed.). Hillsdale, NJ: Lawrence Erlbaum Associates.

I would review the package documentation and reference and if you have further questions, contact the authors of the paper.

Regards,

Marc Schwartz