is sim_slopes supposed to work with brms?

[jacob-long]

is sim_slopes supposed to work with brms?

[jacob-long]

Hi @sag4040, you're not the first one to make that request. I will look into it.

@ebmtnprof, whether brms is supposed to work is a question I hadn't really considered one way or another. It appears that it might "work" (in that it wouldn't fail with errors) if it weren't for a technical glitch. I'd have to do some thinking about whether it makes sense statistically to do simple slopes with a Bayesian model, but empirically it is surely possible. I'll look into making that work in the next major update that I'm working on now, but as a forewarning it may be the case that the model will have to be re-fit several times to get the estimates.

[ebmtnprof]

sorry - it didn't copy the error message. Here it is
Error in terms.default(model) : no terms component nor attribute
In addition: Warning message:
Variable transformations in the model formula detected. Trying to use from global environment instead. This
could cause incorrect results if has been altered since the model was fit. You can manually provide the
data to the "data =" argument.

[sag4040]

Great stuff, Jacob. Thanks for this super helpful package!

I also have a question about the sim_slopes function, but for use with class "glmmTMB". The interact_plot function still works great (warning messages, though), however the sim_slopes function gives me the following error:

Error in UseMethod("summ") :
no applicable method for 'summ' applied to an object of class "glmmTMB"

It would be spectacular if you could point me in the direction of enabling this function for use with glmmTMB models.

All the best,

Scott Graupensperger

[ebmtnprof]

Thanks! I really appreciate the great package :) Cheers, Emily

…

On Sun, Dec 9, 2018 at 3:50 PM Jacob Long ***@***.***> wrote: Hi @sag4040 <https://github.com/sag4040>, you're not the first one to make that request. I will look into it. @ebmtnprof <https://github.com/ebmtnprof>, whether brms is supposed to work is a question I hadn't really considered one way or another. It appears that it might "work" (in that it wouldn't fail with errors) if it weren't for a technical glitch. I'd have to do some thinking about whether it makes sense statistically to do simple slopes with a Bayesian model, but empirically it is surely possible. I'll look into making that work in the next major update that I'm working on now, but as a forewarning it may be the case that the model will have to be re-fit several times to get the estimates. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <https://github.com/jacob-long/jtools/issues/43#issuecomment-445579478>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AkNSOt5IXCUeIlzPt6JY9txdJxQMk_43ks5u3ZPCgaJpZM4XYjLv> .

-- -------------- Emily A. Butler Professor & Graduate Director Family Studies and Human Development College of Agriculture & Life Sciences University of Arizona Tucson, AZ, 85721-0033

[sag40]

Hi Jacob,

Hopefully all is well - I am just wondering if you've had a chance to consider either an update or a potential work-around for using sim_slopes with class "glmmTMB". Unfortunately, my data are count-variables that are zero-inflated, requiring a negative binomial approach.

As always, I appreciate your work on this project. Package developers are the real MVP's of science :)

Scott

[sag40]

Bump^

Hi Jacob,

I am still holding out hope that it may be possible to calculate sim_slopes for glmmTMB objects. If not, I wonder if you've heard of an alternative option to calculate the estimates for the simple slopes at (-1/+1 SD)?

Many thanks!