I have to compare two ICC calculated by the same three judges on the same sample of subjects (presence/absence of a certain physical dysfunction) after and before a training to improve the judges' reliability.
Does any formal testing approach exist?
Thank you
I have performed this kind of analysis before. I prefer to use bootstrap resampling.
For example, you got the 3 judges' measure on 50 subjects before and after training. Select a random sample with replacement (that means the same subject can be selected again), calculate the 2 ICCs and take the difference (after training - before training). Repeat the sampling from your 50 subjects say 1000 times, then you got 1000 difference of ICCs data. Then you can construct a 90% confidence interval by taking the 5th and 95th percentiles. If the lower confidence limit is positive, the after training ICC is significantly better than before. If the upper confidence limit is negative, the after training ICC is significantly worse than before.
You’re right! I will definitely use a bootstrapping approach!
Thank you!
Do you mean a textbook? In case could you please send me the full reference?
Thanks!
I know the test but I cannot figure out how to use it to test the difference between two ICC’s (consider I have three independent evaluators). Could you please detail your strategy?
Thanks!
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis