GAMLJ V. 2.01: Could not find function "extractErrorMessage"

Discuss the jamovi platform, possible improvements, etc.

by rheredia@tamiu.edu » Thu Jan 30, 2020 6:47 pm

Mixed Model (GAMLJ V. 2.01).

Dear Friends,

I just performed a mixed model analyses with Jamovi 1.2.5.0 (GAMLJ V. 2.01) on an iMac with Catalina. It is a 5-way facotr and I have a very unbalanced model ( I know that). I did not get a clean output. It is kind of frozen (grayish) and on the very top of the output, I get this warning in red:

"could not find function "extractErrorMessage."

Again, the output is there, and I know there are a lot of missing values, but I think I supposed to get a clean white screen. I am trying to include a screen shot here, but I am not sure how to included it. I suppose I can email it?



Thanks!

Roberto

rheredia@tamiu.edu

Posts: 12
Joined: Fri Feb 09, 2018 9:41 am
rheredia@tamiu.edu
 
Posts: 16
Joined: Fri Feb 09, 2018 3:41 pm

by jonathon » Thu Jan 30, 2020 11:05 pm

let me attend marcello to this

jonathon
User avatar
jonathon
 
Posts: 1378
Joined: Fri Jan 27, 2017 10:04 am

by mcfanda@gmail.com » Thu Jan 30, 2020 11:17 pm

Dear Roberto
I can help your out if you can send me the .ovm file. Please send it to mcfanda@gmail.com if you have troubles in attaching it here
User avatar
mcfanda@gmail.com
 
Posts: 207
Joined: Thu Mar 23, 2017 9:24 pm

by rheredia@tamiu.edu » Fri Jan 31, 2020 3:34 pm

Dear Jonathan and Marcello,

As always, thank you!!! I already sent the files to Marcello. Thank you so much!

Roberto
rheredia@tamiu.edu
 
Posts: 16
Joined: Fri Feb 09, 2018 3:41 pm

by mcfanda@gmail.com » Fri Jan 31, 2020 6:10 pm

Hi,
here we have two problems. One is a little bug in GAMLj (solved next version, thanks!) that when there's an error in the random component LRT test it fails. If you remove the LRT tests it works. But that is a GAMLj problem. Your model problem is not related to it (but thanks for pointing to the bug). The problems are:
1) The intercepts and the effect of INTERV across TARGET have practically no variance, so their standard errors cannot be computed and the LRT test fails.
2) There's not enough information in the data (due to unbalance) to estimate all the interactions.
Indeed, when you remove the LRT tests it gives you some partial results, but the note says "Some coefficients cannot be....". That means no good.
My suggestion is to "prune out" effects that are not crucial, like the 5-way interactions and see what happens
User avatar
mcfanda@gmail.com
 
Posts: 207
Joined: Thu Mar 23, 2017 9:24 pm

by rheredia@tamiu.edu » Fri Jan 31, 2020 6:22 pm

Dear Marcello,

Thank you!!! I saw the same issue in R and that is why remove the 5-way interaction. Thank you so much for your support!! And anything that I can do to help this wonderful program. BTW, I am in the process on working on a Stats book using Jamovi!
rheredia@tamiu.edu
 
Posts: 16
Joined: Fri Feb 09, 2018 3:41 pm


Return to General