On Wednesday, Jeffrey wrote:
> I have a data set involving 91 subjects. All subjects have been rated
> by the same three raters (R1, R2 & R3) into one of four categories.
His question is:
>
> Given that our rating scale is ordinal, is there a way to
> get STATA to generate a table of individual kappas based on a
> dichotomization of the rating scale at each possible cut-off?
>
If I understand your question correctly, I think you can do this by
replacing each rating with 0/1 based on each cut-off, and then computing
kappa. The code pasted below does this, for a dataset like Jeffrey's,
using a loop.
yours,
May
[email protected]
capture program drop mykap
program define mykap
di as text _n _n _col(3) "Cut" _col(10) "Kappa" ///
_col(25) "Z" _col(36) "Prob>Z"
di in smcl in gr "{hline 41}"
forvalue i=1/3{
preserve
qui replace R1=(R1>`i')
qui replace R2=(R2>`i')
qui replace R3=(R3>`i')
qui kap R*
di as result _col(5) `i' _col(10) %-7.4f r(kappa) ///
_col(25) %-11.2f r(z) _col(35) %-12.4f 1-normprob(r(z))
restore
}
end
clear
set obs 91
gen subject=_n
gen R1=int(uniform()*4)+1
gen R2=int(uniform()*4)+1
gen R3=int(uniform()*4)+1
kap R*
mykap
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
