Skip to content

Fixing variance components in glmmTMB

4 messages · Tempelman, Robert, Ben Bolker

#
Not intentionally/not that I know of.  Is this approach not working 
for you?  Can you provide a reproducible example of what you want to do 
and why it's not working, either here or on Stack Overflow or in the 
glmmTMB issues list?

   cheers
    Ben Bolker
On 2025-01-21 2:33 p.m., Tempelman, Robert wrote:
#
Got it working now...somebody passed onto me this nugget last night:

As of version 1.1.9<https://urldefense.com/v3/__https://cloud.r-project.org/web/packages/glmmTMB/news.html__;!!HXCxUKc!3__nPeuzixezIKH5upUs1PNx5jNLFsdcOqcwQWHN51gD4oFrwbgDeYztNRxll4gyEb7taAg9_BAvDR22KK2sHOPh$>, the argument corresponding to the residual variance (?betad?) is parameterized on the log-SD scale instead of the log-variance scale.   In version 1.1.10, this argument changed from ?betad? to ?betadisp?.

I had "betad" in all of my older code.
#
OK.

   As you point out, both of these are in the NEWS: the change in 
variance scale is admittedly something you just have to know/read about. 
The change from 'betad' to 'betadisp' is something that it might be 
possible to flag with an informative error message ...
On 2025-01-22 10:14 a.m., Tempelman, Robert wrote: