Following methods were used:
20 patients were measured twice (with one device) by 4 different raters, randomly selected from a group of 8. Each of the 8 raters measured 10 patients twice.
Reproducibility is to be determined by calcultating Intraclass Correlation Coefficients (ICC).
Can anyone confirm /comment on following formulas:
a ) Intra rater reproducibility:
ICC= [Var(P) + Var (P*R)] / [Var(P) + Var(M) +Var(P*R) + Var(P*M) + Var(M*R) + Var(E)]
a ) Inter rater reproducibility:
ICC= [Var(P) + Var (C*M)] / [Var(P) + Var(R) +Var(P*R) + Var(P*M) + Var(M*R) + Var(E)]
where Var(P) is the variance due to patients, Var(R) the variance due to raters, Var(M) the variance due to the medical device and Var(E) the variance due to error.
Var (C*M), Var(P*M), Var(P*R) and Var(M*R) are the variances due to the respective interactions.
From Shrout and Fleiss 1979, I cannot easily infer the above mentioned formulas for the ICC. Thanks for any hint in advance!
Hello Scott, I'm not sure if you'll find a definition there, but I sometimes refer to the electronic statistical textbook that is a link on the favourite links page of my website. Thanks
Deborah Hilton Statistics Online
http://sites.google.com/site/deborahhilton/
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics