Crash Modification Factors Research Articles

A Study on the Road Safety Analysis Model: Focused on National Highway Areas in Cheonbuk Province

현재 우리나라의 교통정책은 도로의 신설 확장은 지양하고, 도로의 선형 및 시설을 개량하여 안전성을 증대시키고, 친환경적이며 효율적으로 운영할 수 있는 방향으로 나아가고 있다. 이는 국가 도로사업 중 하나인 제2차 국도 5개년계획('06~'10)이 확장 53건(71%), 개량 22건(29%)인 반면, 제3차 국도 5개년계획('11~'15)은 확장 22건(30%), 개량 50건(70%)로 변화된 것으로 나타나고 있다. 이러한 시설개량위주의 도로사업을 좀 더 효과적으로 추진하기 위해서는 도로의 안전성을 객관적이고 과학적으로 판단하여 사업을 선정하고, 사업에 따른 안전성 향상에 대한 평가가 이루어져야 한다고 판단된다. 본 연구는 이러한 도로별 안전성 분석 및 평가를 위한 모형을 개발하는데 목적이 있다. 본 연구의 주요내용은 미국의 HSM (Highway Safety Manual)을 근간으로 하여 한국실정에 맞게 도로의 안전성을 분석하고 평가할 수 있는 모형을 개발하는 것이다. 모형 정립을 위한 데이터 구축은 전라북도 권역 5개 국도호선을 대상으로 기하구조 요인이 동일하다고 판단되는 구간을 동질성 구간으로 구분하였고, 구분된 1,452개 구간에 대하여 도로 기하구조, 시설물, 교통량, 기상상태, 토지이용 등의 대표값을 수집하였다. 수집된 자료는 교통사고와 각 도로요소의 상관관계 분석을 수행하여 어떠한 요인이 교통사고에 큰 영향을 미치는지 분석하였고, 이를 바탕으로 음이항회귀모형으로 사고모형을 정립하였다. 개발된 모형을 가지고 교통량과 도로구간연장을 이용하여 발생사고건수를 예측하는 안전성능함수와 도로기하구조 및 교통특성 등의 변화에 따라 사고빈도 변화를 결정하는 사고수정계수를 도출하였다. Currently, Korean transportation policies are aiming for increase of safety and environment-friendly and efficient operation, by avoiding construction and expansion of roads, and upgrading road alignments and facilities. This is revealed by that there have been 22 road expansion projects (30%) and 50 road improvement projects (70%) under the 3rd Five-Year Plan for National Highways ('11~'15), while there were 53 road expansion projects (71%) and 22 road improvement projects (29%) under the 2nd Five-Year Plan for National Highways. For more effective road improvement projects, there is a need of choosing projects after an objective and scientific safety assessment of each road, and assessing safety improvement depending on projects. This study is intended to develop a model for this road safety analysis and assessment. The major objective of this study is creating a road safety analysis and assessment model appropriate for Korean society, based on the HSM (Highway Safety Manual) of the U.S. In order to build up data for model development, the sections thought to have identical geometrical structure factors in 5 lines, Cheonbuk province, were divided as homogeneous sections, and representative values of geometric structures, facilities, traffic volume, climate conditions and land usage were collected from the 1,452 sections divided. In order to build up data for model development, the sections thought to have identical geometrical structure factors in 5 lines, Cheonbuk province, were divided as homogeneous sections, and representative values of geometric structures, facilities, traffic volume, climate conditions and land usage were collected from the 1,452 sections divided. The collected data was processed correlation analysis of each road element was implemented to see which factor had a big effect on traffic accidents. On the basis of these results, then, an accident model was established as a negative binomial regression model.Using the developed model, an Crash Modification Factor (CMF) which determines accident frequency changes depending on safety performance function (SPF) predicting the number of accident occurrence through traffic volume and road section expansion, road geometric structure and traffic properties, was extracted.

Safety Effects of Using Short Left-Turn Lanes at Unsignalized Median Openings

The AASHTO Green Book specifically encourages the use of left-turn lanes at median openings on divided roadways to eliminate stopping in through-traffic lanes. However, the lengths for median left-turn lanes recommended by the Green Book are often impractical in urban areas, where the available distance between two adjacent openings is inadequate, and particularly where the volume of left-turning traffic is heavy. Thus, left-turn lanes shorter than the lengths recommended by the Green Book (referred to as short left-turn lanes) are in wide use on urban divided roadways. The objective of this study was to investigate the safety performance of short left-turn lanes at unsignalized median openings. To this end, 6 years of crash data were collected in Houston, Texas, from 52 median left-turn lanes (40 short lanes and 12 lanes with lengths in compliance with Green Book recommendations). A Poisson regression model was developed to relate traffic and geometric attributes to the total number of rear-end, sideswipe, and object–motor vehicle crashes in a left-turn lane. Crash modification factors were calculated for applications to project the crash frequency for a specific change in lane length. Statistical evidence indicates that the difference between actual lane length and Green Book–recommended length has significant effects on crash frequency. However, the increase in crash frequency as a result of short left-turn lanes might be acceptable in cases where engineers must account for traffic and economic and social effects in determining whether a short left-turn lane is appropriate.

Safety Evaluation of Hybrid Main-Line Toll Plazas

Traditional main-line toll plazas on expressways may have both safety and operational challenges. Although many studies have demonstrated the operational and environmental impacts of conversion from traditional toll plazas to a barrier-free system (open road tolling), research that quantifies the safety benefits of new tolling systems is lacking. This study evaluated the safety effectiveness of the conversion from a traditional main-line toll plaza design to a hybrid main-line toll plaza (HMTP) system. An HMTP system combines both open road tolling on the main line and separate traditional toll collection to the side. Various observational before–after studies were applied on 98 main-line toll plazas (two directions) located on approximately 750 mi of toll roads in Florida; 30 of these were upgraded to the HMTP system. The multivariate empirical Bayes method produced the best crash modification factors with low standard errors, and its results indicated that the conversion from traditional main-line toll plaza to HMTP showed an average crash reduction of 47%, 46%, and 54% for total crashes, fatal and injury crashes, and property-damage-only crashes, respectively. The use of an HMTP system also significantly reduced rear-end crashes and lane-change–related crashes by an average of 65% and 55%, respectively. The use of the HMTP system was proved to be an excellent solution to several traffic operations, environmental, and economic problems. The results of this study proved that the safety effectiveness was significantly improved across all locations that were upgraded to an HMTP system.

Transferability of the Highway Safety Manual Freeway Model to the Italian Motorway Network

The currently available release of the AASHTO Highway Safety Manual (HSM) does not include a model for freeways and interchanges. NCHRP Project 17–45 has recently developed specific crash prediction models for this type of infrastructure. Key to making the HSM a standard for road owners and managers worldwide is the transferability of prediction models to different networks. An extensive study conducted on the primary Italian motorway network evaluated the potential issues in application of this methodology to a network characterized by environmental conditions, road characteristics, driver attitudes, and crash reporting systems different from those for which the HSM models were developed. The freeway network considered in this study, approximately 6,700 km long, was represented by 56 freeway sections characterized by an average length of about 12.5 km, covering 700 km of freeway distributed throughout Italy. Four calibration factors were obtained for freeway segments, according to crash severity and type, and two calibration factors were obtained for speed-change lanes, according to crash severity. The results show a good transferability of the analyzed models to the Italian network, especially the freeway models for fatal and injury crashes, although a tendency to underestimate locations of high crash frequency was observed. Some improvements could be made concerning variable calibration factors within the data sets or local calibrations for crash modification factors. Improved localization of the crash data on the Italian road network needs to be highlighted, mainly for speed-change lanes.

Investigating Safety Impact of Edgelines on Narrow, Rural Two-Lane Highways by Empirical Bayes Method

Narrow, rural two-lane highways are mostly characterized by low design features, light traffic volumes with high crash rates, and particularly high fatal crash rates. About 5,000 mi of these highways are administered by the Louisiana Department of Transportation and Development. Run-off-roadway (ROR) crashes are the most common type of crashes on narrow, rural two-lane highways. Because the Manual on Uniform Traffic Control Devices does not require edgelines, edgelines are not found on many such highways because of their low traffic volumes. There are two main concerns for edgeline implementation on narrow two-lane highways: the potential increase in head-on collisions and the added maintenance cost to the already constrained annual maintenance budget. The second part of a study that evaluated the safety impact of edgelines on narrow, rural two-lane highways in Louisiana is presented. On the basis of data collected from 10 locations, the first part of the study proved that edge lines centralized the lateral position of vehicles. This second part of the study evaluated the safety performance before and after the implementation of edgelines on roadway segments selected from all Louisiana districts. By using the empirical Bayes method, the study showed that edgeline implementation significantly reduced expected crash frequencies. Edgeline implementation reduced ROR crashes as well as head-on crashes. The implementation of edgelines benefited primarily male drivers and young drivers. Because of the crash-decreasing trend observed in the 3-year period classified as the after period in this study, the final estimated crash modification factor was 0.85 with a standard deviation of 0.039. The high benefit–cost ratio strongly supports edgeline implementation on narrow, rural two-lane highways in Louisiana.

Analysis of Injury Severity in Crashes on Ramps and at Crossroad Ramp Terminals

Travel along a ramp can present drivers with complex alignment changes and decision points. These changes and complexities can increase the potential for conflict or crash, especially for larger vehicles. A few studies have examined the effect of variables on ramp crash frequency. However, none of these studies considered the effect of variables on crash severity distributions. Relatively little or no information is available about the safety effects of design elements on the severity of ramp crashes. In some cases, countermeasures (such as longitudinal barriers) are implemented to reduce fatal crashes, but the effect of these treatments on less severe crashes is not well understood. New research has been conducted to develop severity distribution functions (SDFs) for ramp segments and signalized and unsignalized crossroad ramp terminals for predicting the proportion of crashes in each severity category as a function of roadway geometric design elements and traffic control features. The SDFs were calibrated with data from California, Maine, and Washington State. The findings from this research show that barrier presence, the number of through lanes, area type, and ramp type influence the proportion of high-severity crashes on ramp segments. At the same time, access point frequency, left-turn operation, the presence of nonramp public street leg, and area type influence the proportion of high-severity crashes at crossroad ramp terminals. These SDFs can be applied along with safety performance functions and crash modification factors to obtain more precise estimates of the safety effects of design decisions.

Crash reconstruction and crash modification factors

This paper addresses the following question: Under what conditions can reconstructed road crashes be used to estimate the effect of a safety-related countermeasure? Results developed by Pearl and his associates are used to draw two main conclusions. First, when one can (1) identify a structural equation describing a type of crash, (2) identify an additional structural equation describing the countermeasure's impact, and (3) estimate the initiating conditions for a set of reconstructed crashes, then a lower bound for a crash modification factor can be estimated by simulating whether or not each of the reconstructed crashes would still have occurred had the countermeasure been present. If the countermeasure's effect is monotonic this bound becomes tight. Second, in situations where it is not possible to reliably identify the structural equations needed for simulation, but where one can (1) identify a set of crash inputs which, when given, make the crash outcome conditionally independent of the countermeasure, and (2) predict how the distribution of these inputs will change in response to the countermeasure, then nonparametric estimation of the countermeasure's crash modification factor is possible. When it is not possible to predict the countermeasure's effect on the conditioning variables it may still be possible to identify constraints or specifications which the countermeasure should satisfy in order to realize a target crash modification.

Four-Lane to Five-Lane Urban Roadway Conversions for Safety

Undivided roadways have consistently exhibited low safety performance, particularly in urban or suburban areas where roadside development is relatively intense. Changing a four-lane undivided road to a divided roadway by either building a boulevard cross-section or installing a physical barrier is a desirable option to improve safety performance of an undivided roadway, but it requires significant resources. This article introduces a crash countermeasure successfully implemented on two different segments of undivided roadways in Louisiana. This crash countermeasure is to change an undivided four-lane roadway to a five-lane roadway with a center lane for left turns by restriping pavement markings without increasing pavement width. Although the five-lane roadway is no longer an acceptable roadway type in Louisiana, the impressive crash reductions on both roadway segments demonstrate it is a feasible solution under constrained conditions. Based on the statistical analysis with 6 years of crash data (3 years before and 3 years after excluding the implementation year), the crash modification factors for both roadways are estimated to be less than 0.5 with a standard deviation less than 0.07. Although it is not surprising to see the biggest crash reduction comes from the rear-end collisions, the other types of collision are also reduced.

A tool for safety evaluations of road improvements

Road safety impact assessments are requested in general, and the directive on road infrastructure safety management makes them compulsory for Member States of the European Union. However, there is no widely used, science-based safety evaluation tool available. We demonstrate a safety evaluation tool called TARVA. It uses EB safety predictions as the basis for selecting locations for implementing road-safety improvements and provides estimates of safety benefits of selected improvements. Comparing different road accident prediction methods, we demonstrate that the most accurate estimates are produced by EB models, followed by simple accident prediction models, the same average number of accidents for every entity and accident record only. Consequently, advanced model-based estimates should be used. Furthermore, we demonstrate regional comparisons that benefit substantially from such tools. Comparisons between districts have revealed significant differences. However, comparisons like these produce useful improvement ideas only after taking into account the differences in road characteristics between areas. Estimates on crash modification factors can be transferred from other countries but their benefit is greatly limited if the number of target accidents is not properly predicted. Our experience suggests that making predictions and evaluations using the same principle and tools will remarkably improve the quality and comparability of safety estimations.

Using multivariate adaptive regression splines (MARS) to develop crash modification factors for urban freeway interchange influence areas

Crash modification factors (CMFs) are used to measure the safety impacts of changes in specific geometric characteristics. Their development has gained much interest following the adoption of CMFs by the recently released Highway Safety Manual (HSM) and SafetyAnalyst tool in the United States. This paper describes a study to develop CMFs for interchange influence areas on urban freeways in the state of Florida. Despite the very different traffic and geometric conditions that exist in interchange influence areas, most previous studies have not separated them from the rest of the freeway system in their analyses. In this study, a promising data mining method known as multivariate adaptive regression splines (MARS) was applied to develop CMFs for median width and inside and outside shoulder widths for “total” and “fatal and injury” (FI) crashes. In addition, CMFs were also developed for the two most frequent crash types, i.e., rear-end and sideswipe. MARS is characterized by its ability to accommodate the nonlinearity in crash predictors and to allow the impact of more than one geometric variable to be simultaneously considered. The methodology further implements crash predictions from the model to identify changes in geometric design features. Four years of crashes from 2007 to 2010 were used in the analysis and the results showed that MARS's prediction capability and goodness-of-fit statistics outperformed those of the negative binomial model. The influential variables identified included the outside and inside shoulder widths, median width, lane width, traffic volume, and shoulder type. It was deduced that a 2-ft increase in the outside and inside shoulders (from 10ft to 12ft) reduces FI crashes by 10% and 33%, respectively. Further, a 42-ft reduction in the median width (from 64ft to 22ft) increases the rear-end, total, and FI crashes by 473%, 263%, and 223%, respectively.

From theory to practice in road safety policy: Understanding risk versus mobility

This paper reviews theoretical issues surrounding transport safety modeling and the implications for road safety policy. The behavioral mechanisms that affect transport safety are typically not considered in safety modeling. These issues are discussed in the context of trade-offs between risk-taking, as perceived by travelers, and other mobility objectives and the attributes associated with them. This is an extension of other theoretical frameworks, such as risk compensation, and attempts to integrate some of the previous frameworks developed over the years. Various examples of behavioral adaptation to specific policies are discussed and linked to the framework. These issues are then discussed in the context of improvements to empirical work in this area and the linkage of theoretical frameworks to crash modeling, in particular the estimation and use of Crash Modification Factors. Conclusions suggest that there are many deficiencies in practice, from estimation of models to choice of effective policies. Progress is being made on the former, while the publication of practical guidance seems to have substantial lags in knowledge.

Safety Evaluation of Converting Traffic Signals from Incandescent to LED Bulbs

Data from 282 signalized intersections in Charlotte, North Carolina, were used to examine the safety effect of converting the signals to composite LED bulbs. An empirical Bayes before-and-after method was used for the evaluation. Since this was a blanket installation by the city of Charlotte, a comparison group of stop-controlled intersections was used to account for possible trends during the study period. Crash modification factors (CMFs) were estimated for three- and four-leg intersections for eight different crash types including crashes at dawn and dusk and in dark conditions. For three-leg intersections, all CMFs were higher than 1.0; this factor indicates a possible increase in crashes due to the LEDs. However, none of these CMFs were statistically different from 1.0 at the .05 significance level. For four-leg intersections, the CMFs associated with rear-end crashes were lower than 1.0 and statistically significant at the .05 level; this finding indicates a reduction in these crash types following the changeover to the LEDs. There was substantial difference between the sites in terms of the effect of the LEDs. The reasons for these differences are not known at this time. Future research should investigate whether LEDs are more or less beneficial depending on the characteristics of the intersection including type of area, sight distance, traffic volume, and phasing scheme.

Evaluation of Safety Performance of Passing Relief Lanes

This paper documents the evaluation of the safety performance of passing relief lanes. A passing relief lane is an intermittently spaced additional lane on a two-lane road that provides drivers the opportunity to pass without having to cross into opposing traffic. The operational benefits are well known, but there is limited substantive evidence on the safety of these lanes. The study was based on data from Michigan, specifically traffic volume and crash history data for seven sites at which passing lanes were implemented within the study period, and for 100 reference sites (without passing lanes) and 231 passing lanes sites that existed throughout the study period. The analysis involved an empirical Bayes before–after evaluation and a comparative, cross-sectional evaluation of safety performance of locations with and without passing lanes. The methodologies were also applied to adjacent nontreated sites 1 mi upstream and downstream of the passing relief lane segment to examine possible spillover effects. From the results, crash modification factors (CMFs) were established for passing lanes in Michigan. These CMFs, which indicated significant safety benefits from passing relief lanes, may be considered for use in locations in other jurisdictions. The results from the cross-sectional and before–after evaluations were generally consistent.

Safety Effectiveness of “Vehicle Entering When Flashing” Signs

This project evaluated the safety effectiveness of “Vehicle Entering When Flashing” (VEWF) intersection conflict warning systems (ICWSs) at stop-controlled intersections. The North Carolina Department of Transportation has utilized this treatment, which consists of vehicle-actuated sign and flasher assemblies placed at or near the intersection to warn motorists on the major or minor road, or both, of vehicles entering the intersection. Although this countermeasure has been used for years in North Carolina, and similar ICWSs have been used in other states, there has been minimal safety evaluation to prove or validate the systems' effectiveness. This study evaluated four categories of VEWF systems. Sites were categorized on the basis of the direction of the alert and placement from the intersection. Crash modification factors (CMFs) were provided for all sites and each category, as well as exclusively for sites with two-lane at two-lane intersections. Empirical Bayes before-and-after techniques were used to overcome the regression to the mean threat. Deployments with major road alerts in advance of the intersection and a combination of both major and minor road alerts were determined to be most effective for two-lane at two-lane stop-controlled intersections with CMFs for total crashes of 0.68 and 0.75, respectively.

Severity Distribution Functions for Freeway Segments

To date, the focus of modeling efforts for freeway safety has been on developing safety prediction functions and crash modification factors, with only limited consideration for crash severity distributions. As a result, relatively little is known about the safety effects of design elements such as lane width, rumble strips, and longitudinal barriers on crash severity. In some cases, countermeasures are implemented with the intent to reduce fatal crashes, but the effect of these treatments on less severe crashes is not well understood. Research was conducted to develop severity distribution functions (SDFs) to predict the proportion of crashes in each severity category as a function of roadway geometric design elements and traffic control features. The SDFs were calibrated with freeway segment data from California, Maine, and Washington State. The findings from this research show that barrier presence, increased traffic volume, increased lane width, and urban area type reduce the proportion of high-severity crashes. At the same time, the presence of rumble strips and horizontal curvature increases the proportion of high-severity crashes. These SDFs can be applied along with safety prediction functions and crash modification factors to obtain more precise estimates of the safety effects of design decisions.

Some Limitations of the Models in the Highway Safety Manual to Predict Run-off-Road Crashes

Crash-prediction models in the current edition of the Highway Safety Manual (HSM) have been developed to predict crash frequency by collision type and severity level for specific types of roadways and sites. Each model is made up of three major components: safety performance functions (SPFs), crash modification factors, and calibration factors. The objective of this study was to identify the limitations of the prediction models in estimating single-vehicle, run-off-road (SVROR) crashes for roadside safety analyses and suggest needed changes and developments. The paper presents a review of the state of the models in HSM and focuses on SPFs. Data from FHWA's safety effects of cross-section design for two-lane roads database were used to gain insight about the characteristics of SVROR crashes and total crashes, and to identify the limitations of the current models in predicting the frequency, type, and severity of SVROR crashes. Three major areas of limitations of SPFs are discussed: (a) assumptions involved in development, (b) variables that are potentially important to roadside design but not considered, and (c) statistical bias and uncertainty of the model equations.

Safety Effects of Horizontal Curve and Grade Combinations on Rural Two-Lane Highways

The safety effects of horizontal curves and grades on highways have been quantified separately, but it is not currently known whether or how the safety performance of horizontal curves interacts with that of grades. Although the first edition of the AASHTO Highway Safety Manual provides crash modification factors for the safety effects of horizontal curvature and percent grade on rural two-lane highways, the manual does not have a method for accounting for the interactions between these effects. In other words, in the Highway Safety Manual procedures for rural two-lane highways, the safety effect of a horizontal curve is the same whether the curve is located on a level roadway, a straight grade, or a vertical curve. Similarly, the safety effect of a straight grade is the same whether it is located on a tangent roadway or horizontal curve. Researchers have always supposed that there are interactions between the safety effects of horizontal and vertical alignments, but these interactions have not been demonstrated in a form useful for safety prediction. The results of research undertaken to quantify the safety effects of five types of horizontal and vertical alignment combinations based on Highway Safety Information System data and crash records from 2003 to 2008 in Washington State are summarized. The outcome is a set of safety pre diction models for fatal and injury and property-damage-only crashes. To present the results in a form suitable for incorporation into the Highway Safety Manual, crash modification factors representing safety performance relative to level tangents were developed from these models for each of the five combinations.

Crash Modification Factors

Crash modification factors (CMFs) are listed in the Highway Safety Manual and other authoritative publications. This information does not allow the reader to distinguish between the predictions of safety effect that can be made confidently and are likely to lead to correct decisions and those that can easily be wrong. Nor can it be known how transferable past research results are to decisions about future actions to be implemented under different circumstances. The conceptual framework described in this paper aims to provide guidance for research about CMFs and for meta-analyses. The central claim is that CMFs are random variables and are not universal constants that apply everywhere at all times. The smaller the standard deviation of a CMF, the more confident the related decision making can be. Therefore, the aim of research into CMFs is to reduce their standard deviations. Ways to do so efficiently are indicated. The requisite theory and equations are provided.

Crash Modification Factors for Changes to Left-Turn Phasing

This study estimated crash modification factors (CMFs) from before–after evaluations of two treatments targeted at reducing left-turn crashes at signalized intersections: (a) changes from permissive to protected–permissive phasing and (b) the implementation of a flashing yellow arrow for permissive left turns. Results of the first evaluation—based on 59 intersections in Toronto, Ontario, Canada, and 12 in North Carolina—indicated a substantial reduction in left-turn opposing through crashes, especially at intersections at which more than one leg was treated, and a small percentage increase in rear-end crashes. For the second evaluation (the implementation of the flashing yellow arrow)—based on data from 51 signalized intersections in Oregon, Washington State, and North Carolina—the results indicated a safety benefit at locations with some kind of permissive left-turn operation before and a disbenefit at locations that had a protected-only operation before. The study estimated the standard deviation of the distribution of the CMF in addition to the conventionally estimated standard error of the mean CMF value. For several CMFs, the standard deviation of the distribution was larger than the standard error of the mean value of the CMF and indicated a substantial variation in the effect of the treatment across different sites. This finding indicates the need for further research into the development of crash modification functions instead of CMFs and for the use of large treatment databases to undertake a more extensive disaggregate analysis of the safety effects. The finding also emphasizes the importance of providing a more explicit consideration of CMF variability in future editions of the Highway Safety Manual.

Evaluation of Two Treatments for Reducing Crashes Related to Traffic Signal Change Intervals

Two related evaluation studies that were conducted under NCHRP Project 17–35 are presented. The research objective was to use the most appropriate analytical methods and the best available data to build on previous research in developing crash modification factors (CMFs) for two treatments for reducing crashes related to traffic signal change intervals: modifying the change interval and installing dynamic signal warning flashers. Three evaluation methods were used as appropriate—the empirical Bayes before–after method, the comparison group before–after method, and cross-sectional multiple regression models. A secondary objective of using cross-sectional models for some evaluations was to examine the comparability of before–after and cross-sectional studies, a subject of topical interest in CMF development. There was a general safety benefit to installing dynamic signal warning flashers, with indications that crash reductions could be obtained overall and for several crash types, including injury, angle, and heavy-vehicle crashes. For the change-interval modification, the before–after study results showed significant reductions (at the 5% level) in total, injury, and rear-end crashes under various scenarios. For both treatments, the results from the cross-sectional analysis were relatively consistent with those from the before–after analysis.