binomial by wilson

[sahoositaram555]

Hi , i have a data set below which i have to use for generating binomial CI by WILSON method and it has to be generated by proc freq.

Please suggest what changes can i make in my code to get the CI with WILSON method?

proc freq data=cases;
by country ;
tables category/ nocum norow binomial(level='Y' cl=WILSON(CORRECT));

weight counts;

output out=out binomial;

run;

Country	infected	total	category	counts
Algeria	10	94	Y	10
Algeria	10	94	N	84
Australia	7	1098	Y	7
Australia	7	1098	N	1091
Austria	8	3025	Y	8
Austria	8	3025	N	3017
Belgium	67	2815	Y	67
Belgium	67	2815	N	2748

[FreelanceReinh]

Hi @sahoositaram555,

Your PROC FREQ code is correct. It does calculate (among other statistics) the Wilson confidence interval (with continuity correction) for the binomial proportion of category 'Y' for each country.

You can use the ODS SELECT statement to restrict the output to the CI if you like. Similarly, you can restrict the output dataset to the confidence limits and the name of the country (and the point estimate, I'd suggest) by using a KEEP= dataset option:

data cases;
input Country :$30. infected total category :$1. counts;
cards;
Algeria 10 94 Y 10
Algeria 10 94 N 84
Australia 7 1098 Y 7
Australia 7 1098 N 1091
Austria 8 3025 Y 8
Austria 8 3025 N 3017
Belgium 67 2815 Y 67
Belgium 67 2815 N 2748
;

ods select binomialCLs;
proc freq data=cases;
by country;
tables category / binomial(level='Y' cl=WILSON(CORRECT));
weight counts;
output out=out(keep=country _: l: u:) binomial;
run;