Snatner Snell 1980 proposed the exact unconditional CI for two independent
proportion difference.
if one group is 0/n... is the CI avaible ?
it is strange for me
as i use proc freq; tables trt*y; exact riskdiff;
(i could not generate exact CI if one group is 0/n )
but if i changed to
proc freq; tables trt * y ; exact or riskdiff; run;
the exact CI for rate difference would be generated in outputs.
is this CI from Santner Snell 1980 ?
