U.S. Department of Transportation

Federal Highway Administration

1200 New Jersey Avenue, SE

Washington, DC 20590

202-366-4000

< Previous | Table of Contents | Next > |

This Desktop Reference provides estimates of the crash reduction that might be expected if a specific countermeasure or group of countermeasures is implemented with respect to intersection crashes, roadway departure and other non-intersection crashes, and pedestrian crashes. The crash reduction estimates are known as Crash Reduction Factors (CRFs). The CRFs presented are the CRF information available to date. In some cases, the CRF is expressed as a Crash Reduction Function.

Where available, the Desktop Reference includes multiple CRFs for the same countermeasure to allow the reader to review the range of potential effectiveness. This Desktop Reference includes CRFs for which the reliability of the estimate is low, or very low. This approach is part of the philosophy of bringing together all the information available to date. (A few CRFs found in the literature were not included in the Desktop Reference. These CRFs were considered to have too large a range or too large a standard error to be meaningful, or the original research did not provide sufficient detail for the CRF to be useful.) The CRFs in this Desktop Reference may be periodically updated as new information becomes available.

A CRF is the percentage crash reduction that might be expected after implementing a given countermeasure. (In some cases, the CRF is negative, i.e. the implementation of a countermeasure is expected to lead to a percentage increase in crashes.) *A CRF should be regarded as a generic estimate of the effectiveness of a countermeasure. The estimate is a useful guide, but it remains necessary to apply engineering judgment and to consider site-specific environmental, traffic volume, traffic mix, geometric, and operational conditions which will affect the safety impact of a countermeasure.* The user must ensure that a countermeasure applies to the particular conditions being considered. The reader is also encouraged to obtain and review the original source documents for more detailed information, and to search databases such as the National Transportation Library (ntlsearch.bts.gov) for information that becomes available after the publication of this Reference.

Traffic engineers and other transportation professionals can use the information contained in this issue brief when asking the following types of question: *Which countermeasures might be considered at the signalized intersection of Maple and Elm streets, an intersection experiencing a high number of total crashes and left-turn crashes? What change in the number of total crashes and left-turn crashes can be expected with the implementation of the various countermeasures?*

In the Tables presented in the Desktop Reference, CRFs are provided in the column "Crash Reduction Factor/Function." The standard error of the CRF is given where available in the column "Std Error." The standard error is the standard deviation of the error in the estimate of the CRF. The true value of the CRF is unknown. The standard error provides a measure of the precision of estimate of the true value of the CRF. A relatively small standard error indicates that a CRF is relatively precisely known. A relatively large standard error indicates that a CRF is not precisely known. The standard error may be used to estimate a confidence interval of the true value of the CRF. (An example of a confidence interval calculation is given below.)

As an example, the CRF for the countermeasure install cameras to detect red-light running for right-angle fatal/injury crashes is 16. The following points should be noted:

- The CRF of 16 means that a 16% reduction in fatal/injury crashes is expected after the installation of red-light running cameras.
- This CRF is bolded which means that a) a rigorous study methodology was used to estimate the CRF, and b) the standard error is relatively small. A CRF which is not bolded indicates that a less rigorous methodology (e.g. a simple before-after study) was used to estimate the CRF, and/or the standard error is large compared with the CRF.
- The standard error for this CRF is 6. Using the standard error, it is possible to calculate the 95% confidence interval for the potential crash reduction that might be achieved by implementing the countermeasure. The 95% confidence interval is ±2 standard errors from the CRF. Therefore, the 95% confidence interval for the installation of red-light running cameras is between 4% and 28% (16 - 2×6 = 4%, and 16 + 2×6 = 28%).
- The reference number is 45 (Persaud et al., as listed in the References at the end of this Desktop Reference).

In some cases, a CRF is given in the form of a function. As an example of a function, consider the countermeasure "Vary truck presence" at 4-leg signalized intersections on rural highways. This function is shown in Table 3. The study was conducted by Bonneson et al.

The function for "Vary truck presence" is:

Where Pt equals percent trucks during the peak hour (average for all intersection movements)

The value of 9 in the function reflects the base condition: 9% trucks at 4-leg signalized intersections during the peak hour on rural highways (average for all intersection movements). If, for example, a practitioner wants to know the safety effect of decreasing the truck presence to 7%, then the resulting CRF value from the function would be 5 (=100×(1-e^{(0.026× (7-9))})). The CRF value of 5 suggests that crash frequency is reduced by about 5% for a 2 percentage point decrease in truck presence (from 9% to 7%).

Twelve Tables of CRFs are provided in this Reference. The Tables are grouped under intersection, roadway departure, and pedestrian crashes, and summarize the information available. The Tables include as much information as is available for each CRF.

The Tables for intersection CRFs contain the following information (where available) for each countermeasure: crash type, crash severity, area type, configuration, control, major road daily traffic volume (vehicles/day), minor road daily traffic volume (vehicles/day), reference, number of intersections observed, crash reduction factor/function, standard error, range, and study type.

The Tables for roadway departure CRFs contain the following information (where available) for each countermeasure: crash type, crash severity, area type, road type, maximum daily traffic volume (vehicles/day), minimum daily traffic volume (vehicles/day), reference, crash reduction factor/function, standard error, range, and study type.

The Tables for pedestrian CRFs contain the following information (where available) for each countermeasure: crash type, crash severity, area type, reference, crash reduction factor/function, standard error, range, and study type.

The following points should be noted:

- The crash severities are: all, fatal/injury (fatal and injury crashes combined), fatal, injury, or property damage only (PDO).
- Where available, the Tables provide existing traffic control information (i.e. the conditions existing before implementation of a countermeasure). The control information for the pre-countermeasure study site may be "no signal," "signal," "stop," or "stop/yield." "No signal" is used when a publication specifies that the intersection was not signalized before the countermeasure was introduced, but does not provide details. (In these cases, the intersection could have yield or stop signs, or no controls at all.) Where the original study is not clear, or omits to give the information, the cell is left blank.
- Road type information (for roadway departure countermeasures) uses the following road types (where available): all, multilane, multilane divided, arterial, highway, or freeway. Where the original study was not clear, or omitted to give the information, the cell is left blank or the study's wording is used.
- In the observed column, a higher number of intersections/sites usually corresponds with a more reliable estimate of the safety effectiveness.
- For some countermeasures, a range of safety effectiveness is provided in the Range Low and Range High columns.
- The study type refers to the methodology used in the CRF study.
- A blank cell means that no information is reported in the source document.
- The following abbreviations appear in the Tables:
- App = Approaches
- Avg = Average
- Config = Configuration
- EB = Empirical Bayes
- Emerg = Emergency
- Max = Maximum
- Min = Minimum
- Obs = Number of observed intersections
- PDO = Property Damage Only
- Ped = Pedestrian
- Ref = Reference
- ROR = Run-Off-Road
- Std Error = Standard Error

- For additional information, please visit the FHWA Office of Safety website safety.fhwa.dot.gov.

< Previous | Table of Contents | Next > |

Page last modified on April 1, 2019