求助：这里有一些医学数据，请各位大侠帮忙处理一下，多谢

shiyiming

表1 治疗前后一般情况比较

组别        人数（男/女）        年龄        DM病程        PDN病程        FBG        HbA1c（%）        BMI（kg/m2）
常规组        42（23/19）        61±11.2        11.5±6.8        5.7±4.8        6.5±1.7        6.9±1.1                         24.8±1.8
中西医组48（28/20）                 60±13.8        11.9±6.3        6.1±5.0        6.6±0.9        7.1±0.6                         23.7±2.2
要求
2组间均数比较用t检验

shiyiming

For example, if you want to compare age in the two groups, the formula may be like the following:
n1 = 42; n2= 48; mean1 = 61; mean2 = 60; s1 = 11.2; s2 = 13.8;
df = n1+n2-2;
s = sqrt[((n1-1)*(s1**2)+(n2-1)*(s2**2))/(n1+n2-2)];
t = (mean1-mean2)/(s*sqrt(1/n1+1/n2));
p =(1-probt(abs(t), df))*2;

assumption:1 no missing; 2.equal std
So, p is the test significance for age difference in the two groups
*****JingJu*****

shiyiming

多谢 大侠

shiyiming

看的不是太懂,你要对哪个变量做T检验? 还有,对于原始数据,不知道怎么读带+-符号的数据,是取均值还是其他?
根据CLT,sample size已经满足正态分布的要求,当然你也可以再做QQ plot或其他正态分布检验程序来证实一下.所以,用T检验室可行的,一般用proc glm/proc univariate/proc ttest.

shiyiming

The key is how much we need to know for an independent T test. I think we have had all we need. It looks like the data was from a meta analysis, wasn't it?
*****JingJu*****