kristinmarshall-NOAA
commented
2 years ago Closing this issue.
You already added the error bars to the plots
Talked to Eric about the random effect values - he confirmed that if you estimate a separate intercept, then the random effects would be centered on zero. He recommended not doing that, though, and just letting the intercept random effect intercepts be what they are.
So great to see the progress with the last few commits: 4a0acf2f9e78930eb3ac58560571193f8e051b02 and 96a2e8df81703eecde74a8d49b2a20943fa796af
A couple of thoughts before we talk tomorrow:
Really cool to see the apparent positive trend in weight at age with the time-varying intercepts. Can you add error bars to the plot to better see where the meaningful differences are?
I'm also wondering about the intercept random effect values all being negative - I would have expected them to be centered around 0, as they appear to be with the catch_month effects. Any ideas why that is?
The cohort effects are all positive, or at least it looks that way to me. Again, I would expect these to be roughly centered on zero. An alternative way to look at cohort effects would be to use the recruitment deviations estimated in the assessment. That could help inform whether there is density dependence happening. Something to think about later on.
Looking forward to chatting tomorrow.