Data/Surveillance

The most important data for evaluating speed management programs and countermeasures are crash data and speed data. Roadway inventory data and other data types (land use, census) may also be helpful to better define the context where speeding occurs. States vary how they define speeding in crash data and speeding may be under-reported (NTSB, 2017). Determining speeding during crash investigations can also be difficult and obtaining consistency may be a challenge even within a jurisdiction using the same crash reporting procedures and definitions (Liu & Chen, 2009). Furthermore, crash data do not account for the rates of exposure of speeding among drivers in crashes and drivers who did not crash. Therefore, jurisdictions are encouraged to supplement crash data with measures of traffic speed.

Speed and speeding data, overlaid with crash data, helps provide a more complete picture of where and when speeding is contributing to crashes. Such locations (or locations with similar risks) should be priorities for speeding interventions. Speed data, due to the strong relationship with safety, can also be used in assessing effectiveness of countermeasures that aim to reduce speeding or speed (as in this guide), and provide an earlier measure of effectiveness than crashes. Citation data should be used cautiously since enforcement allocation may potentially be biased toward high crime and previous high citation areas (Braga et al., 2019; Ingram, 2007). Violation data from speed sensors (including at camera sites or speed display devices) may be useful for tracking changes in compliance at these locations.

A representative sample of traffic speed monitoring data could also be used to assess safety performance in a jurisdiction or on particular types of roads. While Federal regulations in the Manual on Uniform Traffic Control Devices require collection of operating speed data in relation to speed limit-setting (FHWA, 2016) and although States routinely collect speed and speed limit data as part of the Highway Performance Monitoring System (Office of Highway Policy Information, 2016), speed data collection remains inconsistent across jurisdictions, and more detailed and informative speed distribution data have long been viewed by traffic safety experts as helpful in monitoring safety and establishing speed and safety relationships. More disaggregate measures of speed and speed dispersion by short time intervals could be used to better understand how speed and speeding affect safety and factors that may increase or decrease this relationship (TRB, 1998). In Western Australia, for example, road safety officials established an annual speed survey to measure free-flow travel speeds to establish baseline and assess changes in speeds at locations across the road network (not only at treated locations) as a measure of the safety impacts of speed management programs, including automated speed enforcement (Sultana, 2019). The baseline survey was conducted in 2000; 2018 marked the 18th follow-up survey. In addition, the program aimed to assess performance by different contextual factors, including speed limits, road function, regional differences, and traffic volumes. The results have shown improvement in driver compliance with speed limits as well as reductions in 85th percentile speeds due to the efforts in the State.

Speed data may be collected in different ways, depending on the purpose and need. Speed studies that collect data in the field using standard engineering approaches are deemed most valuable for many safety evaluations or site-specific speeding assessments. However, there are many new types of speed data such as probe (moving speeds) and stationary speed data increasingly available (although they may come with a cost from private vendors) that may be useful in assessing speeding and safety risks at a site or across a network. Some jurisdictions may acquire such data for mobility purposes. Users need to understand the characteristics of data collected using different means, and data collected from probes or stationary sensors should probably be validated against other methods by qualified analysts (Li et al., 2019).

NHTSA has developed the Model Minimum Uniform Crash Criteria (currently in the 5th edition; 6th edition expected 2024) to help States improve crash data collection (MMUCC, 2017). Besides collecting elements such as posted speed limit and any driver violations charged, the MMUCC suggests defining “speeding-related” by coding whether each driver involved in the crash was speeding based on verbal or visual evidence.