Accident Data Analysis
All living humans continually bear a certain degree of risk of injury or death. Exponent’s statisticians and data scientists specialize in determining whether a particular activity or product poses an unreasonable risk. Risk estimation involves establishing a reference period and then collecting information about the number of accidents, injuries (or other adverse events) suffered, and the amount of exposure during this period. Because risk is expressed as a rate, measures of both frequency (numerator) and exposure (denominator) are required.
Risk of injury or fatality associated with an activity or a consumer product may be difficult to understand in the abstract. Comparison with the injury or fatality risk associated with familiar products and activities allows the reader to put the risk into context. Information on the frequency of adverse events is usually obtained from observational data collected in reporting systems maintained by government or private sources.
To quantify how machines, vehicles, consumer products, and components behave in their real-world environments, we have developed one of the largest in-house collections of accident and incident data in the country. The in-house accident databases that we maintain include federal traffic crash data:
- Fatality Analysis Reporting System (FARS)
- National Automotive Sampling System (NASS)
- Large Truck Crash Causation Study (LTCCS)
We also have access to Police Accident Report data (PARS) from 23 states.
Other incident databases that we access regularly are the National Fire Incident Reporting System (NFIRS) and the Consumer Product Safety Commission’s (CPSC’s) National Electronic Injury Surveillance System (NEISS), and the National Highway Traffic Safety Administration’s (NHTSA’s) Complaint, Recall and Special Crash Investigation (SCI) databases.
We apply our expertise in accident data analysis in a number of areas, including:
- Motor vehicles
- Consumer products
- Medical devices
- Health and environment
- Engineering reliability
Judging whether the risk of an activity or product is "unreasonable" is typically not as straightforward as merely estimating its risk. In some instances, there may be an absolute standard or risk threshold that cannot be exceeded. When no standard exists, the approach developed at Exponent—and subsequently adopted by such federal agencies as the U.S. Consumer Product Safety Commission—involves comparing the risk and benefits of the activity or product in question to other activities or products judged to be similar in key respects.
Exposure is usually determined from administrative records or sample surveys. Because most accident data analyses are conducted using observational studies, rather than controlled experiments, care must be taken in making comparative risk and benefit judgments. Specifically, before declaring a difference in risk to be significant, one should consider whether any factors other than the factor of primary interest could also have influenced (or "confounded") the outcome and, if so, should adjust for such an effect. This consideration has stimulated the development of increasingly sophisticated statistical approaches to accident data analysis. Exponent is well known for our development of graphical methods for expressing the idea of relative risk and the concept of rare event. These methods help to put risk into perspective.