A method to account for and estimate underreporting in crash frequency research.

⿢Not accounting for crash underreporting may lead to biased estimates.⿢Underreporting models can be used to reduce bias from underreporting.⿢Underreporting model results are compared to random parameters negative binomial.⿢The comparisons indicated that the underreporting models yield better predictions.⿢Underreporting models produce reasonable estimates of crash underreporting.