Skip to content
Prev 15551 / 20628 Next

Power calculations for lmer mixed designs

Dear Phillip,

thank you very much for your good answer - it helped me a lot.

Now that I've been through it a few times, I'm still unclear about a few 
things:
My intercept and effect size is already on the logit scale - the 
intercept is 7.2. But whenever I enter it directly into the qlogis 
function, I'm warned that NaNs are produced. This happens whenever I 
enter anything over 1.0. Does this mean I actually need to enter a 
percentage rather than the absolute value...? Or is the problem an 
entirely different one?
So if I'm seeing this correctly, this way of calculating power is not a 
direct one, but you're rather meant to draw your own conclusions from 
looking at the graphs of the simulations and get a general feel of how 
far off or on your effect size is?
So you can't properly attempt this very way with 4 effects? The author 
implies that you can, but I'm getting the impression it'd be quite 
complicated... I'm not sure running each effect separately would be 
correct.
I'll try those, thanks very much!

Best
Diana