What is the estimate in the model?


Desperately looking for some help in this! I have a model but I am not sure how to actually write the results. It is a gamma log link

The model looks something like:

modeltest <- brm(
  formula = score ~ 1 + group +z.age,
  data = df,
family = Gamma(link = "log"),  seed = 64,chains = 5, control = list(adapt_delta = 0.99),
  iter = 10000,
  cores = 5

and the results something like:

 Population-Level Effects: 
                       Estimate Est.Error    l-95% CI      u-95% CI                 Rhat    Bulk_ESS         Tail_ESS
 Intercept.        -0.91      0.04           -0.99           -0.83               1.01        3260                 62327
 group2            -0.41      0.04           -0.50           -0.32               1.00       44478                 90763
 group3            -0.42      0.05           -0.51           -0.33               1.00.      102421               93703
 z scoreage         0.06      0.02            0.02           0.10                1.00       10103              83437

What is the estimate? Is it beta? I’m not sure how I would write about this.
Am I right in thinking I need to exponentiate the estimate, est.error and CIs? In which case is the estimate now an odds ratio?

Thank you so much for your patience

Over at statsexchange there is a brief summary of how to deal with gamma log links.

Does that help?

There is also a detailed tutorial here using brms

When it comes to writing up the results, it wouldn’t be a bad idea to report a summary of the population-level parameters in a table. But when it comes to explaining the model, you might think in terms of theoretically-meaningful contrasts on the scale of the data. One place to start would be with the conditional_effects() function.

1 Like