标题: test the deviance of two logistic regression model [打印本页] 作者: shiyiming 时间: 2009-4-2 01:58 标题: test the deviance of two logistic regression model 各位，
请问有人知道any syntax of SAS 可以test the deviance of two logistic regression model?
我不想自己运算p-value = 1-pchiq(deviance1, deviance2, 2).
谢谢！
filename mydata 'C:\Documents and Settings\skong\Desktop\mytestSAS.txt';
data mytest;
infile mydata firstobs =2;
input FID $ IID $ PAT MAT SEX PHENOTYPE rs1954187_G PC1 PC2;
run;
proc logistic data=mytest;
model PHENOTYPE (event='1')= rs1954187_G PC1 PC2;
run;
proc logistic data=mytest;
model PHENOTYPE (event='1')= rs1954187_G SEX PC1 PC2 rs1954187_G*SEX ;
test SEX,rs1954187_G;
run;作者: shiyiming 时间: 2009-4-19 06:41 标题: Re: test the deviance of two logistic regression model smaller deviance may be better model because we already count on the degree of freedom. why bothering testing it? sometimes we use deviance to see goodness of fit. that is availabel when you use option scale and aggregate.
i am little confused her: deviance test is any different to likelihood ratio test?