On 6/1/07, emine özgür Bayman <[EMAIL PROTECTED]> wrote:
> Dear R users,
> I'm pretty new on using lmer package. My response is binary and I have fixed
> treatment effect (2 treatments) and random center effect (7 centers). I want
> to test the effect of treatment by fitting 2 models:
>
> Model 1:
HI Emine,
The parentheses specify a variable as random or fixed. You are using
center for random intercepts, but fixed for interactions.
You may want
> mer(cbind( yvect, nvect-yvect) ~ trt + ( 1 | center) + (1 |
> trt:center),
> family = binomial, niter = 25, method = "Laplace", control = l
Dear R users,
I'm pretty new on using lmer package. My response is binary and I have fixed
treatment effect (2 treatments) and random center effect (7 centers). I want
to test the effect of treatment by fitting 2 models:
Model 1: center effect (random) only
Model 2: trt (fixed) + center (random)