The National Academies

NCHRP 17-25 [Completed]

Crash Reduction Factors for Traffic Engineering and ITS Improvements

  Project Data
Funds: $550,000
Research Agency: University of North Carolina - Chapel Hill
Principal Investigator: David Harkey
Effective Date: 8/1/2003
Completion Date: 2/29/2008


Crash reduction factors (also known as accident reduction factors or accident modification factors) provide a computationally simple and quick way of estimating crash reductions. Many states have a set of crash reduction factors that are used for estimating the safety impacts of various types of engineering improvements, encompassing the areas of signing, alignment, channelization, and other traffic engineering solutions. Typically, these factors are computed using before-and-after comparisons, although later research also has suggested the use of cross-sectional comparisons.

Currently, crash reduction factors (CRFs) are used often in the short-term programming process (e.g., an annual review of hazardous locations statewide) to quickly yield a list of improvement sites where the "biggest bang for the buck" is likely. Reliable CRFs could also be used in project development for nonsafety as well as safety-specific projects and could assist agencies in deciding on policies affecting general project design (e.g., context-sensitive design solutions, and traffic calming).

Four impediments exist to using CRFs: (1) While presumed to be based on some type of data analysis, the origins of the factors used in practice are not always clear. Factors that vary from state to state may reflect regional disparities or may indicate a need for updates. (2) CRFs have not been developed for many ITS improvements or other operational strategies. For example, on an urban freeway, the installation of an 8-ft shoulder and the initiation of a safety service patrol both have tangible safety benefits, but CRFs currently exist only for the former. (3) CRFs factors are designed for individual improvements, yet multiple improvements usually occur when an intersection or roadway segment is being rebuilt. (4) existing CRFs often reflect changes in accident experience resulting from improvements at sites experiencing unusually high accident rates. Because of this, the impacts of the improvements tend to be exaggerated (i.e., the phenomenon of regression to the mean).

It should be recognized that CRFs are a tool for quickly estimating the impact of safety improvements. Their strength is that they are relatively quick to use; their weakness is that they are based on limited data. Thus, it is desirable to develop CRFs that consider additional elements (e.g., time of day, weather, and percent of trucks).


The objective of this project is to develop reliable CRFs for traffic engineering, operations, and ITS improvements. Reliable CRFs, at a minimum, meet the following criteria:
  • The CRFs are methodologically and statistically valid. Expert judgment is not a substitute for rigorous analysis. Separate values for CRFs (or a method for adjusting the CRFs) are tabulated that account for various influencing factors such as the highway facility, operating condition, weather, time of day, percentage of truck traffic, and pre-existing crash history as appropriate.
  • The applicability of the CRF is known and documented. For example, some CRFs may denote an impact on crashes only at a specific location whereas other CRFs may affect crashes for an entire stretch of roadway, or some CRFs may apply only to specific accident types or to specific pre-existing conditions (e.g., high percentages of wet weather crashes).
  • The CRFs reflect improvements or combinations of improvements that are of interest to DOTs. Such improvements could, for example, include (i) adding a centerline rumble strip, (ii) modifying a signal in conjunction with adding a left-turn lane or (iii) increasing the frequency of a safety service patrol in concert with improved variable message sign (VMS) signing.
  • The CRFs should represent the different crash categories that reflect the impact of the improvement. Crash categories might include total crashes, severe injury crashes, property damage only crashes, and specific crash types (such as rear end and angle).
  • The CRFs reflect variability. The best estimate of the CRFs, along with some technique that reflects their variability (such as ranges, confidence intervals, standard deviation, or some other technique) should be presented.

Accomplishment of the project objective(s) will require at least the following tasks.

TASKS (1.) Conduct a literature review of ongoing and completed studies that can contribute to the development of reliable CRFs. This literature review may include, but is not limited to (a) the various statistical methods used for computing crash reduction factors, (b) studies of the safety impacts of select countermeasures, and (c) publications that address CRFs directly. (2.) Conduct a survey of the states to identify (a) CRFs currently used in practice and how they are used; (b) improvements or combinations of improvements for which reliable CRFs are most needed (these combinations may be those that states use the most often or those that states feel have the greatest potential for improving safety); (c) published and unpublished studies pertaining to CRFs; and (d) suitable databases that may be used to further develop or validate CRFs. (3.) Conduct follow-up telephone interviews as necessary, to verify the suitability of the databases for use in further developing or validating CRFs. The contractor should identify those to be interviewed based on size and quality of databases, published CRFs, validation efforts for those CRFs (and when those efforts were undertaken), or other considerations. (4.) Evaluate the quality of the available data (quantity, age, format, and other characteristics that may affect data quality) from selected sites within different states. Note that data includes crash data, geometric data, volume data, and other explanatory variables. (5.) Submit an interim report summarizing the findings of Tasks 1 through 4 and including a plan for the development of the reliable CRFs, under Task 7. The plan should specify the methods that will be employed, the data that will be used, what characteristics will be used to modify particular CRFs (e.g., volume and facility type), how the CRFs will be presented to the end user, and how the CRFs will be validated. The plan should also include an outline for the development of preliminary CRFs. These preliminary CRFs refer to select countermeasures (or combinations of countermeasures) that appear promising but for which complete data are not available. Potential examples include, but are not limited to, Intelligent Transportation Systems (ITS) with limited deployment experience, such as variable speed limits. (6.) Meet with the NCHRP panel to review the Task 5 interim report approximately 1 month after its submittal. Submit a revised interim report addressing the panel's review comments. (7.) Execute the approved revised plan. (8.) Submit a final report documenting the entire research effort. The final report shall describe how the project was conducted and include an appendix that presents the crash reduction factors.

Status: The project has been completed. 

Note: The FHWA is conducting a pooled fund study to develop accident modification factors for low cost safety improvements. Information on the status of the pooled fund study may be found at http://www.tfhrc.gov/safety/evaluations.

Product Availability: See below.

NCHRP Research Results Digest 299

NCHRP Report 617: Accident Modification Factors for Traffic Engineering and ITS Improvements.


To create a link to this page, use this URL: http://apps.trb.org/cmsfeed/TRBNetProjectDisplay.asp?ProjectID=451