2
$\begingroup$

I ran the following model and this warning came out, what does this mean?

model:

 mixed_model(fixed = logRT ~ B * P * S * t, 
random = ~1 | Participant, data = Data, family = Gamma(link = "log"))

output:

 Fixed effects:
                                    Estimate Std.Err  z-value   p-value
 (Intercept)                               2.0106     NaN      NaN        NA
 ....

 Optimization:
 method: hybrid EM and quasi-Newton
 converged: TRUE 
 Warning message:
 In sqrt(diag(var_betas)) : NaNs produced
Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

This warning message and the corresponding output indicate that the optimization algorithm was not successful. In particular, for a Gamma mixed it would be better to use the Gamma.fam() implemented in the package. Also, you can have a look in the advice on how to alter the optimization defaults here.

Improve this answer
$\endgroup$
11
  • $\begingroup$ Thank you, I will check the resources. Will the resids_plot function be available for gamma distributions? $\endgroup$
    – CatM
    Commented Sep 7, 2020 at 19:21
  • $\begingroup$ @CatM yes it should work but you need to install the most recent version from GitHub. $\endgroup$
    – Dimitris Rizopoulos
    Commented Sep 7, 2020 at 19:24
  • $\begingroup$ Thank you. Do you mean that I should put family = Gamma.fam()? $\endgroup$
    – CatM
    Commented Sep 7, 2020 at 19:25
  • $\begingroup$ Error in Gamma.fam() : could not find function "Gamma.fam" $\endgroup$
    – CatM
    Commented Sep 7, 2020 at 19:47
  • $\begingroup$ @CatM install the most recent version from GitHub. $\endgroup$
    – Dimitris Rizopoulos
    Commented Sep 7, 2020 at 19:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.