I'm currently using the EMP platform from jmp to determine ICC's for: 30 repeated measures, 1 rater, 10 participants.
Is is possible to deduct CI's for the ICC's?
There are other functions of variance components (some of which may be expressed in terms of standard deviations) that may be nice to have confidence intervals for. The most common one I imagine of interest in addition to the ICC would be the %RR.
The Fit Model output for individual variance components as well as the total variance includes confidence intervals. If you had a traditional gauge R&R design with multiple Parts, Operators, and Replicates in a full factorial design, you could fit a mixed model with Part as a fixed effect, Operator, Part*Operator interaction as random effects, which would also put the Replicate (Residual) in the REML output. In that output, you can get the %RR confidence interval.
Since ICC is considering the ratio of correlated variance components, it's a little bit more complex. But if it can be reported out, that would be great!
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.