kappa calculation and QC in decimal place

[blueskyxyz]

data FatComp;
         input Test Response Count;
         datalines;
1 1  130
1 0  20
0 1  40
0 0  160
;
/*******************************************************************************************
** kappa 
*******************************************************************************************/

data kappa1;
	a=130; b=20; c=40; d=160;
	p0=(a+d)/(a+b+c+d);
	pe=((a+b)*(a+c)+(d+b)*(d+c))/(a+b+c+d)**2;
	se=sqrt((p0*(1-p0)/(1-pe)**2)/(a+b+c+d));
	kappa=(p0-pe)/(1-pe);

	z=PROBIT(1-0.05/2);
	kappa_low=kappa-z*se;
	kappa_high=kappa+z*se;
run;



/*ods trace on;*/
ods output KappaStatistics=Kappa2;
proc freq data=FatComp;
	weight count/zeros;
	tables Test*Response/agree;
/*	test kappa wtkap;*/
/*	exact kappa ;*/
run; 
/*ods trace off;*/

hi sasusers,

please take a look at my code to calculate kappa,  the CI value is a little bit different between kappa1 and kappa2 while the estimate is almost same, so how to modify the code ( get the simple kappa in data step with the formula) to have the same result.

use two methods to double qc the kappa, I want get a 100% equal compare report, BTW, how to get the weight kappa using proc freq in SAS9.4.

Thanks.

Sky.

[blueskyxyz]

/*test data*/
data FatComp;
         input Test Response Count;
         datalines;
1 1  130
1 0  20
0 1  40
0 0  160
;
/*******************************************************************************************
** kappa 
*******************************************************************************************/

data kappa1;
	a=130; b=20; c=40; d=160;
	p0=(a+d)/(a+b+c+d);
	pe=((a+b)*(a+c)+(d+b)*(d+c))/(a+b+c+d)**2;
	se=sqrt((p0*(1-p0)/(1-pe)**2)/(a+b+c+d));
	kappa=(p0-pe)/(1-pe);

	z=PROBIT(1-0.05/2);
	kappa_low=kappa-z*se;
	kappa_high=kappa+z*se;
run;



/*ods trace on;*/
ods output KappaStatistics=Kappa2;
proc freq data=FatComp;
	weight count/zeros;
	tables Test*Response/agree;
/*	test kappa wtkap;*/
/*	exact kappa ;*/
run; 
/*ods trace off;*/

hi sasusers,

please take a look at my code to calculate kappa,  the CI value is a little bit different between kappa1 and kappa2 while the estimate is almost same, so how to modify the code ( get the simple kappa in data step with the formula) to have the same result.

use two methods to double qc the kappa, I want get a 100% equal compare report, BTW, how to get the weight kappa using proc freq in SAS9.4.

Thanks.

Sky.

[blueskyxyz]

the formula is not right, will update later