| |
|
Science Forum Index » Space - Consult Forum » Overall kappa coefficient
Page 1 of 1
|
| Author |
Message |
| Jay Weedon |
 Posted: Thu Dec 28, 2006 3:38 pm |
|
|
|
Guest
|
Happy New Year folks!
For a g x m x m frequency table, SAS PROC FREQ can calculate an
"overall" kappa statistic as a summary of the g separate kappas.
Fleiss's (Statistical Methods for Rates & Proportions, 1981) reference
to this statistic characterizes it as applying when there are g
"independent" estimates of kappa.
I'm not sure what exactly is meant by "independent" here. Does it mean
that the g strata should involve mutually distinct subjects? Or
raters? I've noticed that applications of this method in the
literature tend to be situations where the g strata actually represent
repeated measures by the same raters of the same subjects.
Any light shed will be gratefully received,
JW |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Sun Oct 12, 2008 10:55 am
|
|
