Supplier
DOT/NHTSA
Years Available
2016 to present
Periodicity
Annual
Mode of Collection
Abstraction of administrative/claims data or data from other records; crash reports from law enforcement.
Description
The Department of Transportation, National Highway Traffic Safety Administration (DOT, NHTSA) has collected motor vehicle crash data since the early 1970s to support its mission to reduce motor vehicle crashes, injuries, and deaths on our Nation's highways. The Crash Report Sampling System (CRSS) builds on the retiring, long running National Automotive Sampling System General Estimates System (NASS GES). CRSS is a probability sample of police-reported traffic crashes involving all types of motor vehicles, pedestrians, and cyclists, ranging from property-damage-only crashes to those that result in fatalities. CRSS is used to estimate the overall crash scope and trends, identify highway safety problem areas, measure trends, drive consumer information initiatives, and form the basis for cost and benefit analyses of highway safety initiatives and regulations. NHTSA's crash data collection program consists of CRSS, the Fatality Analysis Reporting System (FARS), the Crash Investigation Sampling System (CISS), Special Crash Investigations (SCI), Non-Traffic Surveillance (NTS), the Crash Injury Research & Engineering Network (CIREN), and special studies conducted to address various safety topics.
Selected Content
Crash related accident, vehicle, and person information.
Population Covered
Target population: all police-reported motor vehicle crashes on trafficways in the United States, excluding territories.
Methodology
CRSS obtains its data from a nationally representative probability sample selected from the estimated 7 million police crash reports that are issued annually. These crashes include those that result in a fatality or injury and those involving property damage only. The police crash reports (PCRs) sample from which CRSS data are coded is a probability sample of police-reported crashes that occurred in the United States. The survey has a multi-stage design. First, the 3,117 counties in the United States were grouped into 707 primary sampling units (PSUs). A PSU in the CRSS is either a county or a group of counties. In the first sampling stage, the 707 PSUs were stratified into 50 strata by the four Census regions, urban/rural, vehicle miles traveled, total number of crashes, total truck miles traveled, and road miles. Then a sequence of nested PSU samples was selected using stratified probability proportional to size sampling method. If necessary, PSU strata were collapsed to ensure that at least two PSUs were selected from each PSU strata. Each PSU sample in the sequence can be used for CRSS data collection. The PSU sample used for the 2016 CRSS data collection has 60 PSUs selected from 24 PSU strata. At the second stage of sampling, first all police jurisdictions (PJs) that issue PCRs for the crashes that occurred within the boundary of the selected PSU is stratified into three PJ strata by the number of PCRs each PJ issued in the past. Then a sample of PJs was selected from each PJ stratum using Pareto sampling. Finally, at the third stage of CRSS sampling, a sample of PCRs was selected from each jurisdiction. To select the PCR sample, first all PCRs issued by the selected PJ for the crashes that occurred within the sampled PSU boundary were stratified into 10 types of accidents. In order for a crash to be eligible for the CRSS sample (1) a PCR must have been completed and reported to the state, (2) the crash must involve at least one motor vehicle travelling on a traffic way, and (3) the result must be property damage, injury, or death. Data collectors visit the selected PJs weekly, sample and copy the PCRs and send them to a central contractor for coding. Trained CRSS coders interpret and code data directly from PCRs into an electronic data file. Approximately 120 data elements are captured. After coding, quality checks are performed and analysis weights are created. A multivariate imputation procedure is then applied on selected variables to perform single imputation of missing values.
Response Rates and Sample Size
The target PCR sample size is about 50,000 every year. In 2016, the CRSS sample was comprised of 47,872 police accident reports (PCRs) of which 47,515 were eligible for inclusion in the final file. These PCRs were collected in 337 police jurisdictions in 53 responding PSUs across the country. Non-response adjustments were applied to all three stages to mitigate the potential non-response bias. In 2017, NHTSA converted 6 of 7 non-responding PSUs to responding PSUs and added one replacement PSU so the PSU sample size became 60. Virtually all sampled PCRs responded.
Interpretation Issues
The CRSS replaced the National Automotive Sampling System (NASS) General Estimates System (GES) in 2016 and has a different sample design and independent sample. In the past, GES fatal crash estimates were significantly lower than NHTSA's FARS fatal crash counts – an indication of GES bias. Thus the 2016 CRSS estimates should not be compared to 2015 and earlier year GES estimates. Although various sources suggest that about half of motor vehicle crashes are not reported to the police, the majority of the unreported crashes involve only minor property damage and no significant personal injury. By restricting attention to police-reported crashes, the CRSS concentrates on those crashes of greatest concern to the highway safety community and the general public.
Limitations
Data year 2016 was the first year of implementation of CRSS. Full site cooperation was established with the 2017 data collection year and estimation protocols are still evolving with this system. Given these constraints, we are unable to project a set availability at this time.
References
- National Center for Statistics and Analysis. (2018, March). Police-reported motor vehicle traffic crashes in 2016. (Research Note Report No. DOT HS 812 501). Washington, DC: National Highway Traffic Safety Administration.
- National Highway Traffic Safety Administration website: Crash Report Sampling System (CRSS). Motor Vehicle Crash Data Collection.
- Zhang, F., Subramanian, R., Chen, C.-L., & Noh, E.Y. (2018. March). Crash Report Sampling System: Design overview, analytic guidance, and FAQs (Report No. DOT HS 812 509). Washington DC: National Highway Traffic Safety Administration.
