c*******o 发帖数: 8869 | 1 用baseline做covariate 去分析 post baseline 变量Y 和change from baseline of
Y (Y-baseline), 另外一个factor 是treatment/vehicle. 现在要测试Y 和Y change
from baseline 在treatment 和vehicle 之间有没有差别. 这两个p value 应该是一样
的把? 我的怎么不一样呢? |
q********i 发帖数: 795 | 2 looks to me you are doing two different tests. The first is a typical ANCOVA
(mean post score among different trt/veh,while controlling basline), but
the second is an ANOVA, coz you are simply testing the mean difference
across different category of trt/veh. |
c*******o 发帖数: 8869 | 3 这么说吧:
for blood pressure"
proc mixed;
model bloodpressure=treatment baseline;
lsmeans treatment/pdiff;
run;
for change of blood pressure from baseline:
proc mixed;
model change =treatment baseline;
lsmeans treatment/pdiff;
run;
I used baseline in both model. should I expect same p value for contrast?
ANCOVA
【在 q********i 的大作中提到】 : looks to me you are doing two different tests. The first is a typical ANCOVA : (mean post score among different trt/veh,while controlling basline), but : the second is an ANOVA, coz you are simply testing the mean difference : across different category of trt/veh.
|
q********i 发帖数: 795 | 4 I dont understand why they should have same p values. You are modeling
different random variables Y and Y_diff which have different mean estimates
【在 c*******o 的大作中提到】 : 这么说吧: : for blood pressure" : proc mixed; : model bloodpressure=treatment baseline; : lsmeans treatment/pdiff; : run; : for change of blood pressure from baseline: : proc mixed; : model change =treatment baseline; : lsmeans treatment/pdiff;
|
a****0 发帖数: 22 | 5 The pvalues should be same, try to derive change again in your data?
【在 c*******o 的大作中提到】 : 这么说吧: : for blood pressure" : proc mixed; : model bloodpressure=treatment baseline; : lsmeans treatment/pdiff; : run; : for change of blood pressure from baseline: : proc mixed; : model change =treatment baseline; : lsmeans treatment/pdiff;
|
q********i 发帖数: 795 | 6 靠,找了个data run 了一下,p值还真是一样,那位老大说一下是为什么啊
【在 a****0 的大作中提到】 : The pvalues should be same, try to derive change again in your data?
|
c*******o 发帖数: 8869 | 7 找到问题了. change from baseline 比blood pressure 少了几个值, 改过来p value
就一样了.
【在 a****0 的大作中提到】 : The pvalues should be same, try to derive change again in your data?
|