I am back again with a few more questions. It will be really helpful to get some guidance. My models have been performing well till I added the random factor (1|followID). Here is the current model and the warning messages.
brm.formula=brmsformula(formula=n|trials(n.act) ~ 1-exp(a)/(1+exp(a))*exp(-exp(b)*exp(-exp(cc)*age)),
a ~ 1 + (1|motherID) + (1|followID),
b ~ 1 + (1|motherID),
cc ~ 1 + predictor2 + predictor4 + predictor5 + predictor3 + (1|motherID),
flist=NULL, nl=T, family=binomial(“identity”)
)
get_prior(brm.formula, data=xdata)
priors=c(
set_prior(“normal(6.5, 1.5)”, class=“b”, coef=“Intercept”, nlpar=“a”),
set_prior(“normal(3.5, 0.6)”, class=“b”, coef=“Intercept”, nlpar=“b”),
set_prior(“normal(0.5, 0.2)”, class=“b”, coef=“Intercept”, nlpar=“cc”),
set_prior(“normal(0,0.2)”, class=“sd”, nlpar=“a”, group=“motherID”),
set_prior(“normal(0,0.2)”, class=“sd”, nlpar=“b”, group=“motherID”),
set_prior(“normal(0,0.2)”, class=“sd”, nlpar=“cc”, group=“motherID”),
set_prior(“normal(0.2,0.4)”, class=“sd”, nlpar=“a”, group=“followID”),
set_prior(“normal(10,1)”, class=“b”, coef=“predictor3”, nlpar=“cc”),
set_prior(“normal(0,1)”, class=“b”, coef=“predictor2M”, nlpar=“cc”),
set_prior(“normal(0,1)”, class=“b”, coef=“predictor4P”, nlpar=“cc”),
set_prior(“normal(0,1)”, class=“b”, coef=“predictor5P”, nlpar=“cc”))
m19 ← brm(formula=brm.formula,
data=xdata,
family=binomial(“identity”),
prior=priors,
control=list(adapt_delta = 0.95, max_treedepth = 15), cores=4,
chains=4, iter=6000, warmup=1000) #increasing the number of iterations
Warning messages:
1: The largest R-hat is 1.66, indicating chains have not mixed.
Running the chains for more iterations may help. See
https://mc-stan.org/misc/warnings.html#r-hat
2: Bulk Effective Samples Size (ESS) is too low, indicating posterior means and medians may be unreliable.
Running the chains for more iterations may help. See
https://mc-stan.org/misc/warnings.html#bulk-ess
3: Tail Effective Samples Size (ESS) is too low, indicating posterior variances and tail quantiles may be unreliable.
Running the chains for more iterations may help. See
https://mc-stan.org/misc/warnings.html#tail-ess
- Could someone please help with specifying priors for followID?
- If I want to plot age vs successes/trials and the fitted model, how can I do that for the above model with all the predictors? The conditional_effects plot doesn’t give successes/trials.
- Finally, if I want to test the significance of this model and get the p value equivalent for each for the predictors, what should I do?
Sorry, I am new to fitting Bayesian models and have little clue about model fitting and inference.