p********a
发帖数: 5352 | 1 ☆─────────────────────────────────────☆
lilyrong (ATLAS) 于 (Thu Aug 7 14:32:57 2008) 提到:
我该如和解释？我在研究高血压（yes or no) 在给不同药的小组 (药A and 药B组）5
年的变化，我的结果是A明显比B组的人在1-5年的高血压少很多。没给药前，2组
baseline 是matched 的。所以我用的是repeated measures, linear mixed model。
请问我该怎么回答？万分感谢。
For binary outcomes (e.g. hypertension), what type of mixed model was
utilized? Is this a linear model / GEE model, please explain?
☆─────────────────────────────────────☆
vincentsg (vincentsg) 于 (Thu Aug 7 14:49:24 2008) 提到:
i don | j*****e
发帖数: 182 | 2 Because the outcome is binary, you should use generalized linear mixed model
. The term "mixed" means you have random terms in your model. This is caused
by your design(repeated measurement). Do it either in genmod or glimmix.
The parameter estimation method is GEE.GEE is not a model. It is an
estimation method, different from the traditional likelihood method. |
|
