Common Data Elements between the Large Truck Crash Causation Study Investigations and Commercially Available Onboard Monitoring Systems

Abstract

At the heart of traffic safety is the identification of factors that lead to crashes. With this knowledge, interventions can be developed to mitigate or prevent these factors from occurring in the future. Post hoc reconstructions of crashes (e.g., the Large Truck Crash Causation Study; LTCCS) and naturalistic driving studies have provided information on crash genesis. However, there is another source of driving data that is currently untapped. Today, numerous commercial vehicle fleets, transit fleets, and personal vehicles use onboard safety monitoring (OBSM) systems to monitor and improve driving behavior. Data from these video-based OBSM systems could be used by researchers to learn more about crash genesis and address some of the limitations inherent in post-crash reconstruction. This study created a data directory of common data elements in the LTCCS, commercially available, video-based OBSM systems, and other public sources to be used together to provide researchers with more valid and reliable information on crash genesis. Researchers used the LTCCS codebook as the structure of the new data directory. The LTCCS variables were analyzed by a trained researcher, who determined whether the variable could be collected with an OBSM system or through related information by using one of three responses: yes, no, and maybe. Lytx™ and SmartDrive were used as comparison OBSM system vendors to determine the kind and types of data they can capture. Other related sources such as Police Accident Reports (PARs) were used as possible sources of information for each variable when relevant. If a variable was labeled as “MAYBE,” the conditions under which the variable could be captured by a video-based OBSM system or through related information were outlined. Analysis determined that approximately half of all 802 variables in the LTCCS codebook could be captured using video-based OBSM systems and related information. In addition, another almost 30% of the variables were labeled as “MAYBE.” An analysis of the data collected in the actual LTCCS revealed that only 75% of the crashes had data and 48% of the variables had the option to be coded as “unknown.” Based on the results in the current report, it appears that the use of OBSM systems and other data sources could yield a similar amount of data as that obtained by data analysts in the LTCCS.