Crash modification factors, when used properly, can help transportation engineers identify and apply the most appropriate countermeasures for increasing roadway safety.

Transportation funding is a limited resource with many competing needs—maintaining existing infrastructure, constructing new facilities, and operating transportation systems safely. To help stretch their budgets available for highway safety, State and local transportation agencies work to identify and implement the optimal combination of countermeasures to achieve the greatest benefits. Equipped with the right tools, transportation engineers can make informed decisions to save money and reduce crashes.

Crash modification factors (CMFs) are tools that, when applied correctly, can help to identify the expected safety impacts of installing various countermeasures to reduce crashes. CMFs are multiplicative factors used to estimate the number of crashes after implementing a given countermeasure at a specific site. Combined with crash cost data and project cost information, CMFs can help transportation engineers compare the benefit-to-cost ratio of multiple countermeasures and then choose the most appropriate CMF for a given situation.

The Federal Highway Administration (FHWA) is leading a concerted effort to develop information on CMFs and make it available to State and local agencies to assist with highway safety planning. The CMF Clearinghouse, a free online data--base introduced in 2009 and accessible at www.cmfclearinghouse.org, details the varying quality and reliability of CMFs available to transportation professionals. (For more information, see "The CMF Clearinghouse: A Handy Safety Tool" in the November/December 2010 issue of PUBLIC ROADS.)

Today's challenge is to help State and local transportation agencies identify appropriate CMFs and then apply them effectively. What follows is a primer on how to choose the most applicable CMFs and apply them successfully to help reduce crashes and save lives.

Selecting CMFs

Selecting an appropriate CMF is similar to choosing the right tool for the job. In some cases, a tool that may not be perfect will still work well enough to get the job done. For example, a crescent wrench will tighten bolts even though an exact size wrench might be the ideal tool. Although the exact size may be ideal, the crescent wrench is adjustable and versatile and therefore can serve the purpose. The same can be true with CMFs. Even if a CMF is not a perfect match for the situation to which it is applied, it may work well enough to provide a reasonable estimation of the countermeasure's effect.

"You have to do the best you can with the information available," says Shawn Troy, a safety evaluation engineer with the North Carolina Department of Transportation (NCDOT). For example, NCDOT recently needed a CMF for total crashes for installing in-lane pavement markings as a supplemental measure to enhance guidance on a multilane facility. Based on a query of the CMF Clearinghouse, the engineers identified three CMFs for "mark pavement with supplementary warning." Only one dealt with total crashes, indicating an expected reduction of 6 percent. In addition, the CMF was general (that is, no specific indication of applicable roadway type, area type, or number of lanes) and was not rated due to a lack of supporting information. However, this was the only information available for reference, so the NCDOT engineers made the decision to apply this general and unrated CMF until a more rigorous study is completed.

In other cases, using an improper tool may do more harm than good. For example, using a hammer to tighten bolts may not be effective at meeting the objective. Similarly, applying a CMF that does not fit the specific situation may not allow a reasonable estimate of the countermeasure's safety effectiveness. Selecting an inappropriate CMF may lead to one of two outcomes—the change in crashes will be over- or underestimated. "The risk of applying an inappropriate CMF is that you may end up selecting a less cost-effective treatment," says Troy.

Three main considerations are necessary to assure appropriate selection of CMFs: the availability of relevant CMFs, the applicability of available CMFs, and the quality of applicable CMFs.

Availability. The availability of a CMF that applies to a specific situation depends on whether research has been conducted to determine the safety effects of a particular countermeasure or combination of countermeasures, and whether researchers have documented it. The CMF Clearinghouse contains more than 2,900 CMFs and receives quarterly updates to include the latest research. The American Association of State Highway and Transportation Officials' Highway Safety Manual (HSM) also provides CMFs for various countermeasures but contains only a subset of those found in the clearinghouse. The HSM focuses on higher quality CMFs for specific roadway characteristics and countermeasures.

Applicability. Once a transportation professional determines that one or more CMFs exist for a specific countermeasure, the next step is to determine which CMF is the right one for the job, that is, which is most applicable. Applicability depends on how closely the CMF represents the situation to which it will be applied. For example, if a transportation engineer is trying to decide whether to install chevrons along a horizontal curve on a rural, two-lane road, the engineer will first identify the potential safety effectiveness of installing chevrons. The engineer then will evaluate the potentially applicable CMFs, eliminating any that are not appropriate for the situation. In this example, the engineer will eliminate the CMFs for combination treatments, even if they include chevrons, because this situation includes only the chevrons. Likewise, the engineer will eliminate CMFs for urban areas, since the situation at hand is in a rural setting. Any options that have a significantly different traffic volume thus also will be eliminated.

After dismissing the less appropriate CMFs, the engineer still may have multiple CMFs to choose from to estimate the effectiveness of the chevrons. Final selection of a CMF will depend on the objective of the analysis. If the objective is to estimate the reduction in fatal and injury crashes, then the engineer will choose the CMF noted for that specific crash type and severity. If the engineer chooses to use a CMF outside the range of applicability, the safety effect may be over- or underestimated.

"Several variables can be used to match a CMF to the scenario at hand," says Daniel Carter, senior engineering research associate with the Highway Safety Research Center at the University of North Carolina. "Transportation engineers can use treatment type, roadway type, area type, segment or intersection geometry, segment or intersection traffic control, traffic volume, and originating State to help them determine the best CMF to use."

Crash modification functions (CMFunctions) can help calculate point estimates of a CMF over a range of values for a given variable. For example, a CMF may change as traffic volume increases or decreases. In this case, a CMFunction would estimate the effectiveness of the countermeasure using the traffic volumes associated with the site, thereby resulting in a more applicable CMF. As the state of the knowledge advances, CMFunctions will replace CMFs because a single point estimate may not accurately describe the safety effects of a given countermeasure.

Quality. Often a search for applicable CMFs generates multiple CMFs for the same countermeasure. To help make a selection, an engineer will examine the quality of each CMF. According to Craig Lyon, a principal with Persaud and Lyon, Inc. and an international expert on developing CMFs, "not all CMFs are created equal." In other words, the quality of a CMF can vary greatly depending on several factors associated with the process of developing the CMF. The primary factors that determine the quality of a CMF are the study design, sample size, standard error, potential bias, and data source.

Both the HSM and the CMF Clearinghouse provide some indication of the quality of each CMF. The CMF Clearinghouse provides a star rating for each based on a scale of 1 to 5, where 5 indicates the highest quality. The most reliable CMFs in the HSM are indicated with a bold font.

In a case where two or more applicable CMFs for the same countermeasure have the same star rating but different CMF values, the engineer would use other details to determine whether one is more applicable to the situation in question. For example, an engineer in Virginia is considering the use of high-friction surface treatments to address wet-weather crashes on a horizontal curve. The engineer identifies two CMFs with an identical star rating, but one was developed based on data from Kansas and the other was developed based on data from North Carolina. The more appropriate selection might be the CMF from North Carolina because the climate, topography, and other characteristics are more similar to Virginia than those in Kansas.

Applying CMFs

Just as the application of an appropriate CMF can influence the decision to implement a particular project, the misapplication of a CMF can lead to misinformed decisions. Three main factors need to be considered when applying CMFs: (1) how to estimate the number of expected crashes without treatment (that is, to what number is the CMF applied), (2) how to apply CMFs by type and severity, and (3) how to apply multiple CMFs if multiple treatments are to be included in the same project.

Estimating expected crashes without treatment. Before applying CMFs, transportation engineers first need to estimate the expected safety performance without any treatment or countermeasures. The CMF then is applied to that number to estimate the expected crashes with treatment. The HSM presents several methods for estimating the expected safety performance of a roadway or intersection including the empirical Bayes method, which combines observed information from the site of interest with information from similar sites to estimate the expected crashes without treatment.

For example, consider a four-legged signalized intersection with 12 reported crashes in the past year. To estimate the expected crashes without treatment, an engineer will use the safety performance functions in the HSM to compute the predicted number of crashes. The predicted number of crashes (for this example, 7.7 crashes in a single year) is estimated based on the crash history at several nearby sites with similar characteristics (four-legged signalized intersections with similar geometry and traffic volumes). The empirical Bayes method then combines the observed (12) and predicted (7.7) crashes using a weighted average to estimate the expected crashes without treatment. Assuming a weight of 0.75, which favors the observed crashes, the expected number of crashes is: 0.75(12) + 0.25(7.7) = 10.9. If the engineer simply had used the observed crashes as an estimate of the expected crashes without treatment, he or she might have overestimated the potential effectiveness of a treatment.

Applying CMFs by type and severity. CMFs may apply to total crashes or to target crash types and severities. In many circumstances, estimating the change in crashes by type and severity is useful; however, transportation engineers only can use this approach when CMFs exist for the specific crash types and severities in question. The crash type associated with a CMF defines the crashes for which the related countermeasure is targeted. For example, a CMF for shoulder rumble strips may be applicable to run-off-road crashes. Crash severity is defined by the most severe outcome of the crash. For example, a CMF might apply to crashes resulting in fatalities, injuries, or property damage only.

Applying a CMF for a specific crash type or severity to other crash types and severities may lead to skewed estimates because a countermeasure may reduce certain crash types and severities but increase other crash types and severities. For example, roundabouts are expected to reduce fatal and injury crashes because they eliminate crossing-path collisions; however, there is the potential for an increase in property-damage-only crashes (such as rear-end and sideswipe crashes), particularly when drivers are unfamiliar with driving through a roundabout.

Source: CMF Clearinghouse.

In the example provided earlier, an engineer was trying to estimate the potential safety effects of installing chevrons along a horizontal curve on a rural, two-lane road. If the engineer is interested in both the potential change in total crashes and the impact on crash type and severity, he or she will apply a CMF for total crashes and then also apply CMFs for the targeted crash types (head-on, run-off-road, and sideswipe) and severities (fatal, serious injury, and minor injury) separately. In each case, with the respective CMF, the engineer will use the general equation for estimating the number of crashes after treatment: Expected Crashes After Treatment = CMF x Expected Crashes Without Treatment.

Applying multiple CMFs. CMFs are available for many countermeasures, but most are related to only a single countermeasure. In real-world scenarios, transportation agencies commonly install more than one countermeasure. Engineers then must ask, "What is the safety effect of the combined treatments?" John Milton, chair of the Highway Safety Performance Committee of the Transportation Research Board (TRB) and director of risk management for the Washington State Department of Transportation, says, "At present, the answer is not so clear. However, research is underway to determine the best solution for calculating the safety effects of a combination of countermeasures."

Currently, the common practice is to assume that CMFs are multiplicative. In other words, each successive countermeasure will achieve an additional benefit when implemented in combination with other countermeasures. This is the current method presented in the HSM and in the CMF Clearinghouse. However, transportation agencies also are using other methods. Based on a recent survey of State departments of transportation conducted under National Cooperative Highway Research Program Project 17-25, "Crash Reduction Factors for Traffic Engineering and ITS Improvements," other methods include applying the CMF for the single countermeasure expected to achieve the greatest reduction, applying CMFs separately by crash type and summing them to get a project-level effect, and applying engineering judgment based on a review of crash patterns.

Assuming CMFs are multiplicative means assuming that the full benefit of each countermeasure is expected. However, this is unlikely when two countermeasures address the same crash types. For example, assuming the CMF for shoulder widening is 0.89 and the CMF for installing shoulder rumble strips is 0.85, the combined effect using the multiplicative rule is: 0.89 x 0.85 = 0.76. While shoulder widening is likely to reach its full safety benefit and rumble strips will add an additional benefit, it is not likely that the rumble strips will achieve the full expected reduction beyond the shoulder widening because they address similar crash types.

Regardless of the method employed, engineering judgment is required when combining multiple CMFs. If multiple countermeasures target the same crash type, engineers may consider applying only the most effective CMF, or use the multiplicative method for combining CMFs and reduce the CMFs for additional treatments by a percentage. The HSM presents a method for combining multiple CMFs that assumes they are multiplicative and that the user may estimate the combined effect of multiple treatments as the product of the respective CMFs.

"Regardless of the method an agency chooses to handle the application of multiple CMFs, it is important to apply the method consistently throughout the agency to ensure a fair comparison of projects," says TRB's Milton.

Summary

To help prevent misapplication of CMFs, which can lead to over- or underestimating potential benefits and subsequently making misinformed decisions, each agency needs to apply CMFs consistently for its projects.

Identifying and selecting applicable CMFs is the first step toward achieving consistency. Applicable CMFs are those that closely match the situation at hand. Factors that influence the applicability and selection of a CMF include the treatment type, roadway type, area type, segment or intersection geometry, segment or intersection traffic control, traffic volume, and originating State. If multiple applicable CMFs are available for a given treatment, the quality of a CMF is another factor engineers can consider to differentiate the results.

Applying CMFs also requires engineers to consider multiple factors for consistency. The three main factors are estimating expected crashes, applying CMFs by type and severity, and applying multiple CMFs. The key to applying CMFs is to apply them only to situations for which they were developed.

Future Research

While the HSM, CMF Clearinghouse, and other related resources provide nearly 3,000 CMFs for various safety strategies, several knowledge gaps and opportunities still exist. In fact, the American Association of State Highway and Transportation Officials' Subcommittee on Safety Management task force overseeing the HSM has identified the development of additional CMFs as a high-priority research need for the second edition of the HSM. There are several planned and ongoing efforts to develop CMFs, including FHWA pooled-fund studies, National Cooperative Highway Research Program projects, and research at the Turner-Fairbank Highway Research Center. The CMF Clearinghouse also provides a link for users to submit current CMF needs, which FHWA uses to guide future CMF research projects.

Moving forward with the development of new CMFs, researchers are focusing on producing the most reliable CMFs using appropriate methods. In the meantime, transportation engineers must weigh the strengths and weaknesses of the various methods and existing CMFs developed from those methods. In December 2010, FHWA published A Guide to Developing Quality Crash Modification Factors (FHWA-SA-10-032) to help highway safety analysts identify appropriate methods for developing CMFs. The guide provides an overview of each method and the data required to employ a given method. A safety analyst can use the guide to identify data requirements for various methods and select the most appropriate method based on the data available to them. The guide also outlines the strengths and weaknesses of various methods, which practitioners may use to determine the relative quality of CMFs.

Future research also will focus on developing guidance for applying multiple CMFs. The primary issue is that current methods for combining multiple CMFs may overestimate the potential safety effects, particularly if the treatments target similar crash types. Currently, FHWA is exploring existing methods for applying multiple CMFs to better define the issues and propose alternative methods.

"The key moving forward," says Monique Evans, director of the FHWA Office of Safety Research and Development, "is to ensure that future efforts target knowledge gaps and develop high-quality CMFs through rigorous analysis."

References

1. American Association of State Highway and Transportation Officials. Highway Safety Manual, 1st Edition, Washington, DC, 2010.
2. Council, F., E. Zaloshnja, T. Miller, and B. Persaud. Crash Cost Estimates by Maximum Police-Reported Injury Severity Within Selected Crash Geometries (FHWA-HRT-05-051). FHWA, McLean, VA, 2005.
3. Crash Modification Factors Clearinghouse. U.S. Department of Transportation and the University of North Carolina Highway Safety Research Center. Available online at: www.cmfclearinghouse.org (Accessed April 4, 2011).
4. Gross, F., C. Lyon, and B. Persaud. A Guide to Developing Quality Crash Modification Factors (FHWA-SA-10-032). FHWA, Washington, DC, 2010.

Frank Gross, Ph.D., P.E., is a highway safety engineer with Vanasse Hangen Brustlin, Inc. (VHB). He specializes in the evaluation of safety strategies, development of CMFs, and road safety assessments. He earned a Ph.D. in civil engineering from The Pennsylvania State University with a graduate minor in statistics.

Karen Yunk, P.E., is the Highway Safety Improvement Program implementation manager with FHWA's Office of Safety Programs. She has an M.S. in civil engineering from Rutgers, the State University of New Jersey.

For more information, contact Frank Gross at 919–834–3972 x5602 or fgross@VHB.com, or Karen Yunk at 609–637–4207 or karen.yunk@dot.gov.