01739560

Component

https://doi.org/10.1177/0361198120916469

Unsafe driving behaviors, driver limitations, and conditions that lead to a crash are usually referred to as driver errors. Even though driver errors are widely cited as a critical reason for crash occurrence in crash reports and safety literature, the discussion on their consequences is limited. This study aims to quantify the effect of driver errors on crash injury severity. To assist this investigation, driver errors were categorized as sequential events in a driving task. Possible combinations of driver error categories were created and ranked based on statistical dependences between error combinations and injury severity levels. Binary logit models were then developed to show that typical variables used to model injury severity such as driver characteristics, roadway characteristics, environmental factors, and crash characteristics are inadequate to explain driver errors, especially the complicated ones. Next, ordinal probit models were applied to quantify the effect of driver errors on injury severity for rural crashes. Superior model performance is observed when driver error combinations were modeled along with typical crash variables to predict the injury outcome. Modeling results also illustrate that more severe crashes tend to occur when the driver makes multiple mistakes. Therefore, incorporating driver errors in crash injury severity prediction not only improves prediction accuracy but also enhances our understanding of what error(s) may lead to more severe injuries so that safety interventions can be recommended accordingly.

© National Academy of Sciences: Transportation Research Board 2020.

English

pp 452-464

2020-5

Transportation Research Record: Journal of the Transportation Research Board

Web

References (51)

Crash data; Crash injury research; Driver errors; Injury severity; Logits; Mathematical prediction; Probits

Highways; Safety and Human Factors

TRIS, TRB, ATRI

May 13 2020 3:06PM