Full Bayes Poisson gamma, Poisson lognormal, and zero inflated random effects models: Comparing the precision of crash frequency estimates

► Random effects, zero inflated, and zero inflated random effects models of crash data are compared. ► The models are compared in terms of goodness-of-fit measures. ► They are also in terms of precision of posterior crash frequency estimates. ► By including random effects, zero inflated models degenerate to their non-zero inflated counterparts.