calculate ICC for consistency in JASP
Hi everyone!
Is it possible to calculate the intraclass correlation coefficient for consistency (rather than absolute agreement) in JASP? if so, how is it done? I have a repeated-measures design in which participants perform the same task twice in two different sessions two weeks apart and I would like to assess the test-retest reliability (over RTs measures for the two sessions) using ICC.
Thank you!
Comments
Hi Pmeng,
Let me start by saying that I'm not a member of the development team. But, after having a quick look at the documentation for the function I presume they use, it would seem that only the absolute agreement version is currently implemented. I would suggest that this could be a good feature request.
See
How to Request a Feature or Report a Bug in JASP - JASP - Free and User-Friendly Statistical Software
Best,
Tarandeep
Hi,
I apologize for the delay in response. Even though I did not construct the analysis in JASP myself, I think the ICC analysis offers what you re looking for when the ICC(3,1) is reported. According to the underlying R-package, the ICC-types are calculated according to Shrout And Fleiss (1979). Looking into that source, I find that the formula for the ICC(3,1) is the formula for consistency instead of absolute agreement. The formula for absolute agreement adds a term in the denominator (see McGraw and Wong, 1996). See also the psych function definition: https://rdrr.io/cran/psych/man/ICC.html;
Julius