How can I found the appropriate family to fit my data using brms?

Interfaces brms
,
1

Hello, everyone. I have a set of data from dose-response effect (without saturation point), but I don’t know how to fit it with appropriate family.

I have values that follows a log distribution, however, there are some negative values, and the Log family doesn’t work.

I’m trying to use 2 set of priors since from data the curves from one group are diff (one group starts reacting at the lowest concentration, while the other only at the highest). I have different bees and whitin each bee I record different glomeruli

fit3 <- brm(Signal ~ mo(OConc_ord) * Group + (1|Bee:GloTag), 
            data = df,
            prior   = c(prior_nurses,
                        prior_foragers),
            chains  = 4,
            iter    = 2000,
            warmup  = 1000,
            cores   = 4,
            seed    = 123,
            sample_prior = TRUE,
            family = gaussian(log))

code_to_run_your_model(if_applicable)

Thanks!!!

:D

2

Choosing an appropriate response distribution in a regression generally requires one or more of:

  • clear domain knowledge that furnishes an empirical or theoretical expectation for the response distribution
  • data exploration in the actual data you’ve collected
  • iterative model building, criticism, and refinement

To begin, can you describe your prior or theoretical expectations for how the response should be distributed? Is there a theoretical expectation the outcome should be strictly positive? If so, what accounts for negative values in your sample?

And can you provide a plot of the response against the OConc_ord, maybe either for one representative Group or with separate colors or panels for different Group values?

2 Likes