What prompted this wishlist request?
There is a lot of variation in the stage of cancer, pain, and level of allergies, so a class value within 10, typically 0 to 6, is often used.
When comparing these detection methods and measurement techniques, the simple kappa coefficient does not assess ordinal weight and would weight incorrectly identifying cancer stages 0 and 1 in the same way as incorrectly identifying stages 0 and 4.
Therefore, I am performing bivariate analysis between ordinal scales, but I am having trouble calculating weighted kappa in the current version of JMP.
I know there are add-ins, but since there are many beginners, it would be helpful if they were easy to use.
What would you like to see improved?
I would like to be able to calculate weighted kappa from bivariate analyses of ordinal scales.
Correlation coefficients can also be depicted somehow using the heat map in Graph Builder, but since it is necessary to adjust the axes, it would be convenient to be able to easily depict a class correlation diagram.
Why is this idea important?
Measurements with many classes are plentiful.
Cancer, dementia, pain, severity, proficiency, phase, etc.
Class correlation diagrams are also powerful tools when explaining things to users or superiors.
Furthermore, if weighted kappa values can be calculated, it will be possible to directly compare correlation diagrams using weighted kappa values.
Thank you for your wish. We will discuss this internally and will come back to you.
