F8
发帖数: 348 | 1 dataset with
a binary outcome y/n
4 different studies (treated as an 4-level categorical variable) and 3
treatment for each study
in a logistic model
found significant interaction of trt*study
Now try to consider random effect with study level covariate
As I know that nlmixed can only handle two random effects, in this case
Is there a way to fit a random effect model by treating study as a random
effect? |
c********h
发帖数: 330 | 2 你是说computationally?
try lme4 package in R
glmer(y ~ trt + (trt|study), family = "binomial") |
F8
发帖数: 348 | 3 我的问题是这种情况下用 nlmixed 怎么来 fit random study effect
glmer 相当于sas glimmix ？
能解释一下什么意思吗
哪个是fixed/random effect in this model
study 是categorical 还是 numeric
【在 c********h 的大作中提到】
: 你是说computationally?
: try lme4 package in R
: glmer(y ~ trt + (trt|study), family = "binomial")
|
d********t
发帖数: 837 | 4 If all you want is study as random effect and treatment as fixed effect ,
that is literally the most straightforward case for glimmix. Not sure what
you are going for .
dataset with a binary outcome y/n4 different studies (treated as an 4-level
categorical ........
【在 F8 的大作中提到】
: dataset with
: a binary outcome y/n
: 4 different studies (treated as an 4-level categorical variable) and 3
: treatment for each study
: in a logistic model
: found significant interaction of trt*study
: Now try to consider random effect with study level covariate
: As I know that nlmixed can only handle two random effects, in this case
: Is there a way to fit a random effect model by treating study as a random
: effect?
|
F8
发帖数: 348 | 5 I did it actually
the results from glimmix with or without random effects in this case are
identical to the
4-th decimal which is hard to understand
level
【在 d********t 的大作中提到】
: If all you want is study as random effect and treatment as fixed effect ,
: that is literally the most straightforward case for glimmix. Not sure what
: you are going for .
:
: dataset with a binary outcome y/n4 different studies (treated as an 4-level
: categorical ........
|
