Hello,
I am part of a study that is looking at 4 clinicians abilility to make a continuous variable measurement with 5 different techniques. We have a gold standard that is a direct measurement of the parameter of interest.
I have generated a delta (the actual measurement minus the clinicians measurement) for each data point.
I am trying to figure out how to first look at each clinicians intraobserver variability using each of the 5 techniques. I was thinking if I just using distribution and then getting a p value for the mean delta (compared to mean of 0 if measurement was perfect) to assess each clinician's 'skill' at each technique.
is that apropriate?
Then I want to get a sense of intraobserver variability/agreement (are these the same thing?) for each clinician using each technique.
What test would I run to get that agreement (ICC, CCC?) and would i be able to do that in JMP?
then I was thinking we could evaluate each technique across the 4 clincians to get the interobserver relaibility for each technique.
is that doable with JMP? Is that also ICC or CCC?
Finally, I was thinking about pooling all the clinicians together per each tehcnique and then measuring the agreeement between the 5 different techniques.
I would also like to compare each of tho technuiques against our gold standard ( the direct measurement) to get a sense of the 'best' technique.
is that doable in JMP?
Sorry this is a lot, any help is appreciated.
My stats knowlege is pretty slim
