Calculating SEM in test-retest study (ICC)

#1
Hi!

I am involved in the writing of a thesis, where two raters are running a test-retest study to test the intra- and inter-rater reliability of a clinical test. Here the Standard Error of Measurement (SEM) needs to be reported, and the following formula should be used: SEM=SD x √(1-ICC). I have read similar research papers, but I see that they have used different SDs. In SPSS (ICC analysis) the section "Item statistics" provides the SD for both rater 1 and rater 2, and one under "Scale Statistics". I also see that some research papers have been operating with SDpool. Which SD should be used when calculating SEM?