In N() operator, provide a way to specify different network-level predictors for different terms.

[krivit]

Motivating example: we want GWESP with decay common to all networks in the sample but coefficient varying, say, as a function of net size. Specifying N(~gwesp(), ~log(n)) will have net size effect for both sample and decay. How do we want to specify the UI?

[krivit]

UI already allows this to be done: offset(N(~gwesp(), ~log(n)), 4) with offset.coef=0 will "lock" the slope for the decay parameter at 0. TODO: Document this somewhere.

[cmarcum]

would the interpretation of the statistic effectively be a grand mean over all networks and all layers? I can add the documentation: I've been meaning to do that to clean up some typos anyway

On Tue, Jul 20, 2021, 8:22 AM Pavel N. Krivitsky @.***> wrote:

UI already allows this to be done: offset(N(~gwesp(), ~log(n)), 4) with offset.coef=0 will "lock" the slope for the decay parameter at 0. TODO: Document this somewhere.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <statnet/issue-mover#27>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAI377F7ZL32AZ5K7SZ25STTYVS7NANCNFSM5AQVXTFA .

[krivit]

I guess it could be viewed as a weighted mean, but generally it's a pooled MLE (e.g., https://pubmed.ncbi.nlm.nih.gov/32547745/). That said, I would suggest holding off on updating the documentation, since it's going to be ported to the new documentation system that should let you do it in Rmarkdown.

[krivit]

And thanks for offering!

[krivit]

@cmarcum , the documentation has been ported, in case you've been waiting.