Hi
I'm trying to calculate a measure of agreement, but have a fairly unusual
design.
I have 28 raters, 12 subjects and 3 categories. For some subjects (3),
there are some missing ratings (1 or 2 for each).
I have used the 'kappa' command in Stata which seems to handle this
design, but this doesn't give a confidence interval.
A paper was published in Biometrics in 1986 (42, 883-893), in which Gross
presents measures of agreement and confidence intervals, for such designs
where the number of raters is large relative to the number of categories
and subjects (albeit with a higher number of raters). Does anyone happen
to know if Stata/any other package has a command for this, or know of
another appropriate method for handling this design (which provides a
confidence interval)?
Many thanks
Melissa
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|