y***o
发帖数: 416 | 1 统计专业问题请教,谢谢大家
在研究过程中有一个难题,请大家帮忙
有两组人群 <= 18 岁, > 18 岁. 两种药物效果如下：
new medicine traditional medicine incremental
counts % of recovery counts % of recovery
<= 18 10,000 4.20% 5,000 3.10% 1.10%
> 18 30,000 3.5% 15,000 2.10% 1.40%
现在的难题是如何对INCREMENTAL做SIGNIFICANT TEST
证明1.1% 和1.4% 是SIGNIFICANTLY DIFFERENT
也就是证明新药相对于旧药对成人更有效
谢谢大家！ | o*******w
发帖数: 2310 | 2 binomial confident interval?
(p1-p2)+-1.96*s.e
if 0 not inside the interval, then, significant | c****y
发帖数: 94 | 3 用proc logistic. 我反正没事帮你做了第一个。
proc logistic data=x;
class medicine;
model response(event="1")=medicine;
run;
pvalue is 0.001
Odds ratio 1.37 95% C.I. 1.136 to 1.653.
It concludes there is a statistical significant increase. However, whether
this increase is significant in pratical sense? It is your call. |
|
