Crash modification factors for high friction surface treatment on horizontal curves of two-lane highways: A combined propensity scores matching and empirical Bayes before-after approach.

Horizontal curves are locations that, as a result of the changing alignment, may be a contributing factor in roadway departure crashes. One low-cost countermeasure to mitigate crashes at these locations is the installation of the high friction surface treatment (HFST), which increases roadway friction and is intended to help keep drivers on the roadway when traversing a horizontal curve. This treatment has been implemented at numerous curves in Pennsylvania, but the overall safety effectiveness is not known. The purpose of this study is to estimate a suite of Crash Modification Factors (CMFs) for HFST applied to curve sections of undivided two-lane roadways. A novel combination of the empirical Bayes observational before-after study design and propensity score matching was used to estimate CMFs for multiple crash types, crash severities, and roadway settings (urban and rural). Propensity score matching was implemented to identify the most appropriate reference group to use within the empirical Bayes methodology. The results indicate that the installation of HFST is associated with a statistically significant decrease in all crash types and severities considered.

Copula-based bivariate count data regression models for simultaneous estimation of crash counts based on severity and number of vehicles.

Statistical models of crash frequency typically apply univariate regression models to estimate total crash frequency or crash counts by various categories. However, a possible correlation between the dependent variables or unobserved variables associated with the dependent variables is not considered when univariate models are used to estimate categorized crash counts-such as different severity levels or numbers of vehicles involved. This may lead to inefficient parameter estimates compared to multivariate models that directly consider these correlations. This paper compares the results obtained from univariate negative binomial regression models of property-damage only (PDO) and fatal plus injury (FI) crash frequencies to models using traditional bivariate and copula-based bivariate negative binomial regression models. A similar comparison was made using models for the expected crash frequency of single- (SV) and multi-vehicle (MV) crashes. The models were estimated using two-lane, two-way rural highway segment-level data from an engineering district in Pennsylvania. The results show that all bivariate negative binomial models (with or without copulas) outperformed the corresponding univariate negative binomial models for PDO and FI, as well as SV and MV, crashes. Second, the statistical association of various traffic and roadway/roadside features with PDO and FI, as well as SV and MV crashes, were not the same relative to their corresponding relationships in the univariate models. The bivariate negative binomial model with normal copula outperformed all other models based on the goodness-of-fit statistics. The results suggest that copula-based bivariate negative binomial regression models may be a valuable alternative for univariate models when simultaneously modeling two disaggregate levels of crash counts.

Examining the impacts of crash data aggregation on SPF estimation.

The American Association of State Highway and Transportation Officials' Highway Safety Manual (HSM) includes a collection of safety performance functions (SPFs) or statistical models to estimate the expected crash frequency of roadway segments, intersections, and interchanges. These models are applied in several steps of the safety management process, including to screen the road network for opportunities to improve safety and to evaluate the performance of safety countermeasure deployments. The SPFs in the HSM are generally estimated using negative binomial regression modeling. In some instances, they are estimated using annual crash frequency and site-specific (e.g., traffic volume) data, while in other instances they are estimated using aggregate crash frequency and site-specific data. This paper explores the differences that result from estimating SPFs using aggregate versus disaggregate data using the same methods as those used to estimate the SPFs in the HSM. A synthetic dataset was first used to conduct these comparisons - these data were generated in a manner that is consistent with the properties of the negative binomial distribution. Then, an observational dataset from Pennsylvania was used to compare the SPFs from both aggregate and disaggregate data. The results show that SPFs estimated using the panel (disaggregate) data and aggregated data provide similar model coefficients, although some differences may sometimes arise. However, the overdispersion parameter obtained using each dataset can differ significantly. These differences result in systematic biases in calculations of expected crash frequency when Empirical Bayes adjustments are applied, which - as the paper demonstrates - could lead to different outcomes in a network screening exercise. Overall, these results reveal that aggregating crash data might result in biased SPF outputs and lead to inconsistent Empirical Bayes adjustments.

Estimating safety performance functions for two-lane rural roads using an alternative functional form for traffic volume.

Statistical models of expected crash frequency are referred to as Safety Performance Functions (SPFs) in the first edition of the American Association of State Highway and Transportation Officials' Highway Safety Manual (HSM). The SPFs in the HSM specify expected annual crash frequencies as a function of various roadway and roadside features, with the most important predictor variable being traffic volume, which serves as a measure of vehicle exposure to crashes. Traffic volumes are typically measured using the average annual daily traffic and are incorporated into the SPFs using a natural logarithm transformation. This specification suggests that the relationship between expected crash frequency and traffic volume increases non-linearly with a constant elasticity over the range of observed values. While researchers concur that the relationship between expected crash frequencies and traffic volume is non-linear, further exploration of the functional form of this relationship may offer additional insights concerning the association between safety performance and vehicle exposure. This paper proposes an alternative functional form for the traffic volume variable in SPFs that allows for different elasticities between traffic volume and expected crash frequency within different traffic volume ranges, while preserving the same general non-linear relationship in existing HSM SPFs. Although other forms-like the Hoerl function-have been proposed in the literature, the proposed model allows for natural breakpoints in the traffic volume for which roadway or geometric features might have varying effects on low- or high-volume roads. The proposed functional form was applied to SPFs developed for two-lane rural roadways in Pennsylvania. Comparisons with SPFs developed using the traditional and Hoerl functional forms suggest that this proposed functional form offers an improved fit and predictive performance, and thus might be considered for the development of future SPFs.

Empirical Bayes before-after evaluation of horizontal curve warning pavement markings on two-lane rural highways in Pennsylvania.

Roadway departure crashes contribute to a large proportion of fatal and injury crashes in the United States. These crash types are more likely to occur along horizontal curve sections of a roadway. Countermeasures that prevent vehicles from departing the roadway is one method to mitigate roadway departure crashes. Pennsylvania has deployed on-pavement horizontal curve warning markings in advance of horizontal curves on two-lane rural highways as a roadway departure crash reduction strategy. This study used an Empirical Bayes (EB) before-after study design to evaluate the safety effects of the horizontal curve warning pavement markings. A total of 263 treatment sites and more than 21,000 reference sites were included in the evaluation. Crash modification factors were developed for total, fatal plus injury, run-off-road, nighttime, nighttime run-off-road, and nighttime fatal plus injury crashes. The point estimates for each of these crashes ranged from 0.65 to 0.77 - the results were statistically significant for total and fatal plus injury crashes at the 95th-percentile confidence level.

Crash modification factors for adaptive traffic signal control: An Empirical Bayes before-after study.

Adaptive traffic signal control (ATSC) is a novel traffic management system that is often deployed at high-volume intersections in order to mitigate traffic congestion and improve travel time reliability. While past studies have demonstrated its operational effectiveness, relatively few have focused on safety performance. Those that have tend to suffer from limitations including small sample sizes, insufficient study designs, or the lack of consideration of potential temporal and corridor effects after ATSC installation. Furthermore, results from previous studies are mixed: while many studies point to a safety improvement, more recent studies seem to indicate that ATSC systems might increase crash frequency. In light of this, a comprehensive Empirical Bayes (EB) before-after observational study was conducted using ATSC data collected throughout Pennsylvania. Crash modification factors (CMFs) were estimated based on the following different case scenarios: crash severity levels and crash types (total, fatal and injury, rear-end, and angle crashes); intersection locations (all intersections and intersections along corridors only); and, intersection configurations (3-leg and 4-leg). Temporal trends for intersection-level CMFs were examined using annual crash data in the after period. Corridor-level CMFs were also developed to quantify changes in safety performance along corridors with ATSC installed. The results suggest that ATSC is associated with a nominal increase in total and angle crashes, and an expected decrease in fatal plus injury crashes and rear-end crashes. However, the results were not statistically significant. The safety effect estimates are similar when considering intersection locations and configurations. In addition, the temporal trend analysis indicates that the safety effectiveness does not vary annually in the after period, suggesting no obvious novelty effect associated with ATSC. Finally, the magnitude of the corridor-level CMFs are slightly lower than the intersection-level CMFs, except for rear-end crashes.

Incorporating Bayesian methods into the propensity score matching framework: A no-treatment effect safety analysis.

The propensity score matching method has been used to estimate safety countermeasure (treatment) effects from observational crash data. Within the counterfactual framework, propensity score matching is used to balance the covariates between treatment and control groups. Recent studies in traffic safety research have demonstrated the strength of this method in reducing the bias caused by treatment site selection. However, several general issues associated with safety effect estimates may still influence the effectiveness and robustness of this method. In the present study, Bayesian methods were integrated into the propensity score matching method. Bayesian models are known for their ability to capture heterogeneity and modeling uncertainty. This may help mitigate unobserved variable effects in the roadway and crash data. Furthermore, the sampling-based algorithm used for Bayesian estimation yields more consistent estimates in small region analysis than estimates from frequentist modeling. In this study, a dataset that was used to evaluate the safety effects of the dual application of shoulder and centerline rumble strips on two-lane rural highways was acquired. Only data from the before treatment period were used in a no-treatment effect analysis in order to compare the results of a Bayesian propensity score analysis to a frequentist propensity score analysis. Because no treatment was applied during the analysis period, it was assumed that there would be no treatment effect, or a crash modification factor equal to 1.0. The Bayesian propensity score matching method nominally outperformed the frequentist propensity score matching method in the largest sample and produced near-identical results in the medium sample, but neither method closely approximated the assumed, true crash modification factor in the small sample analysis. A simulation study is recommended to further study the effects of sample size and confounding factors when comparing the Bayesian and frequentist propensity score matching methods.

Application of a model-based recursive partitioning algorithm to predict crash frequency.

Count regression models have been applied widely in traffic safety research to estimate expected crash frequencies on road segments. Data mining algorithms, such as classification and regression trees, have recently been introduced into the field to overcome some of the assumptions associated with statistical models. However, these data-driven algorithms usually provide non-parametric output, making it difficult to draw statistical inference or to evaluate how independent variables are associated with expected crash frequencies. In this paper, the model-based recursive partitioning (MOB) algorithm is applied in a crash frequency application. The algorithm incorporates the concept of recursive partitioning data in tree models and develops user-defined statistical models as outputs. The objective of this paper is to explore the potential of the MOB algorithm as a methodological alternative to parametric modeling methods in crash frequency analysis. To accomplish the objective, a standard negative binomial (NB) regression model, a NB model developed using the MOB algorithm, adjusted NB models which incorporate variables identified by the MOB algorithm, and a random parameters NB model are compared using 8 years of data collected from two-lane rural highways in Pennsylvania. The models are compared in terms of data fitness, sign and magnitude of statistical association between the independent and dependent variables, and predictive power. The results show that the MOB-NB model yields better data fitness than other NB models, and provides similar performance to the RPNB model, suggesting that the MOB-NB model may be capturing unobserved heterogeneity by dividing the data into subgroups. The presence of a passing zone and posted speed limit are two covariates identified by the MOB algorithm that differentiate variable effects among subgroups. In addition, the MOB-NB model provides the highest prediction accuracy based on the training and test data sets, although the difference among models is small. The comparison results reveal that the MOB algorithm is a promising alternative to identify covariates, evaluate variable associations and instability, and make predictions in a crash frequency context.

Evaluating the predictive power of an SPF for two-lane rural roads with random parameters on out-of-sample observations.

Negative binomial (NB) regression is among the most common statistical modeling methods used to model crash frequencies due to its simple functional form and ability to handle over-dispersion commonly found in crash data. However, a drawback of this approach is that regression parameters are assumed to be the same across observations, which could contribute to biased parameter estimates. To alleviate this concern, the random parameters negative binomial (RPNB) model was recently proposed, which allows regression parameters to differ across observations following some known distribution. The resulting coefficients should be less biased, and thus the RPNB approach is believed to provide a more accurate relationship between independent variables and expected crash frequency. However, the prediction accuracy of the RPNB model relative to the standard NB model has not been thoroughly evaluated, particularly with respect to out-of-sample observations for which unique random parameters cannot be estimated. In this paper, the predictive power of the RPNB and NB models are examined using two-lane rural highway data from three engineering Districts in Pennsylvania. Multiple evaluation metrics are applied-root-mean-square error (RMSE) and mean absolute error (MAE), coefficients from calibration functions and cumulative residual (CURE) plots-to assess each model type. The results show that the RPNB model outperforms the NB model when applied to within sample observations (i.e., those used to estimate the model) by making use of the observation-specific coefficients. However, the predictive power of the RPNB model appears to be similar to or slightly less precise than the traditional NB model when applied to out-of-sample observations. Since the RPNB model is estimated using a simulation-based approach, sensitivity tests were also performed to see how the parameter estimates change with the number of Halton draws used to perform the simulation. For the sample sizes used in this paper, the estimates were fairly insensitive when more than 50 Halton draws were used. The findings suggest that the RPNB model is more reliable when applied to the same set of sites that were used to estimate the model but might not be as robust as the traditional NB model when applied to other sites.

Safety and operational impacts of setting speed limits below engineering recommendations.

This study quantifies the operational and safety impacts of setting posted speed limits below engineering recommendations using field data from rural roads in Montana. Vehicle operating speeds and historical crash data were collected at multiple sites with posted speed limits set equal to engineering recommendations and sites with posted speed limits set lower than engineering recommendations. Linear, quantile and logistic regression models were estimated to predict mean operating speed, 85th percentile operating speed and speed limit compliance, respectively, as a function of various roadway characteristics and level of speed enforcement. The Empirical-Bayes before-after approach was also used to develop crash modification factors (CMFs) that describe the expected change in total and fatal + injury crash frequency when setting posted speed limits lower than engineering recommendations. Because safety data were collected over a long time period, temporal adjustments were incorporated to account for yearly changes in crash reporting, traffic characteristics and other variables. The results revealed that speed limit compliance worsened as the difference between the engineering recommended and posted speed limits increased. The presence of verified heavy police enforcement reduced both mean and 85th-percentile operating speeds by approximately 4 mph and increased speed limit compliance. The safety analysis found a statistically significant reduction in total, fatal + injury, and property damage only (PDO) crash frequency at locations with posted speed limits set 5 mph lower than engineering recommendations. Locations with posted speed limits set 10 mph lower than engineering recommendations experienced a decrease in total and PDO crash frequency, but an increase in fatal + injury crash frequency. The safety effects of setting speed limits 15 to 25 mph lower than engineering recommendations were less clear, as the results were not statistically significant, likely due to the small sample of sites included in the evaluation. Overall, the results suggest that setting posted speed limits 5 mph lower than the engineering recommended practice may result in operating speeds that are more consistent with the posted speed limits and overall safety benefits.

Safety performance functions for horizontal curves and tangents on two lane, two way rural roads.

Horizontal curves on two-way, two-lane rural roads pose critical safety concerns. Accurate prediction of safety performance at these locations is vital to properly allocate resources as a part of any safety management process. The current method of predicting safety performance on horizontal curves relies on the application of a safety performance function (SPF) developed using only tangent sections and adjusting this value using a crash modification factor (CMF). However, this process inherently assumes that safety performance on curves and tangent sections share the same general functional relationships with variables included in the SPF, notably traffic volumes and segment length, even though research suggests otherwise. In light of this, the goal of this paper is to systematically study the relationship between safety performance and traffic volumes on horizontal curves of two-lane, two-way rural roads and to compare this to the safety performance of tangent sections. The propensity scores-potential outcomes framework is used to help ensure similarity between tangent and curve sections considered in the study, while mixed-effects negative binomial regression is used to quantify safety performance. The results reveal that safety performance on horizontal curves differs significantly from that on tangent sections with respect to both traffic volumes and segment length. Significant differences were also found between the safety performance on tangents and curves relative to other roadway features. These results suggest that curve-specific SPFs should be considered in the next edition of the Highway Safety Manual.

Investigation of flashing and intensity characteristics for vehicle-mounted warning beacons.

Reducing the potential for crashes involving front line service workers and passing vehicles is important for increasing worker safety in work zones and similar locations. Flashing yellow warning beacons are often used to protect, delineate, and provide visual information to drivers within and approaching work zones. A nighttime field study using simulated workers, with and without reflective vests, present outside trucks was conducted to evaluate the effects of different warning beacon intensities and flash frequencies. Interactions between intensity and flash frequency were also analyzed. This study determined that intensitiesof 25/2.5 cd and 150/15 cd (peak/trough intensity) provided the farthest detection distances of the simulated worker. Mean detection distances in response to a flash frequency of 1 Hz were not statistically different from those in response to 4 Hz flashing. Simulated workers wearing reflective vests were seen the farthest distances away from the trucks for all combinations of intensity and flash frequency.

Development of regionalized SPFs for two-lane rural roads in Pennsylvania.

The American Association of State Highway and Transportation Officials' Highway Safety Manual (HSM) contains safety performance functions (SPFs) to predict annual crash frequencies for several roadway types. When applying these SPFs in a jurisdiction whose data were not used to develop the SPF, a calibration factor can be applied to adjust the expected crash frequency estimate to statewide or local conditions. Alternatively, the HSM suggests that transportation agencies may develop their own SPFs in lieu of applying the calibration factor to the HSM SPFs. However, the HSM does not provide guidance on the appropriate level of regionalization that should be adopted for either method, even though safety performance may vary considerably within a state. In light of this, the present study considers the development of local or regionalized SPFs for two-lane rural highways within the Commonwealth of Pennsylvania. Three regionalization levels were considered: statewide, engineering district and individual counties. The expected crash frequency for each level of regionalization was compared to the reported crash frequency over an eight-year analysis period. The results indicate that district-level SPFs with county-level adjustment factors provide better predictive accuracy than the development of a statewide SPF or application of the HSM-calibrated SPF. The findings suggest that there are significant differences in safety performance across engineering districts within Pennsylvania. As such, other state transportation agencies developing SPFs or using calibration factors may also consider how variations across jurisdictions will affect predicted crash frequencies.

Causal inference framework for generalizable safety effect estimates.

This study integrates a causal inference framework to the Empirical Bayes (EB) before-after method to develop generalizable safety effect estimates (i.e., crash modification factor (CMF)). The method considers approaches to estimate the average treatment effect for the treated (ATT), average treatment effect for the untreated (ATU), and average treatment effect (ATE). The current EB method is shown to estimate ATT while ATE is what is typically desired in traffic safety research. Modifications to the current EB method to estimate ATU and ATE are provided. The method is then applied to a dataset with a "no-treatment" scenario where the treatments were: 1) randomly selected and 2) selected based on crash history. Given the "no-treatment" outcome, it is known that the CMFs should have a value of 1 in order to be considered accurate. The standard negative binomial and mixed effects negative binomial regression models were applied in the analysis. It was found that, of the two regression methods, the ATE CMFs developed using the standard negative binomial were the most accurate. Finally, potential sources of bias in the EB method are discussed.

A method to account for and estimate underreporting in crash frequency research.

Underreporting is a well-known issue in crash frequency research. However, statistical methods that can account for underreporting have received little attention in the published literature. This paper compares results from underreporting models to models that account for unobserved heterogeneity. The difference in the elasticities between the negative binomial underreporting model and random parameters negative binomial models, which accounts for unobserved heterogeneity in crash frequency models, are used as the basis for comparison. The paper also includes a comparison of the predicted number of unreported PDO crashes based on the negative binomial underreporting model with crashes that were reported to police but were not considered reportable to PennDOT to assess the ability of the underreporting models to predict non-reportable crashes. The data used in this study included 21,340 segments of two-lane rural highways that are owned and maintained by PennDOT. Reported accident frequencies over an eight year period (2005-2012) were included in the sample, producing a total of 170,468 segment-years of data. The results indicate that if a variable impacts both the true accident frequency and the probability of accidents being reported, statistical modeling methods that ignore underreporting produce biased regression coefficients. The magnitude of the bias in the present study (based on elasticities) ranged from 0.00-16.79%. If the variable affects the true accident frequency, but not the probability of accidents being reported, the results from the negative binomial underreporting models are consistent with analysis methods that do not account for underreporting.

Safety evaluation of continuous green T intersections: A propensity scores-genetic matching-potential outcomes approach.

The continuous green T intersection is characterized by a channelized left-turn movement from the minor street approach onto the major street, along with a continuous through movement on the major street. The continuous flow through movement is not controlled by the three-phase traffic signal that is used to separate all other movements at the intersection. Rather, the continuous through movement typically has a green through arrow indicator to inform drivers that they do not have to stop. Past research has consistently shown that there are operational and environmental benefits to implementing this intersection form at three-leg locations, when compared to a conventional signalized intersection. These benefits include reduced delay, fuel consumption, and emissions. The safety effects of the conventional green T intersection are less clear. Past research has been limited to small sample sizes, or utilized only statistical comparisons reported crashes to evaluate the safety performance relative to similar intersection types. The present study overcomes past safety research evaluations by using a propensity scores-potential outcomes framework, with genetic matching, to compare the safety performance of the continuous green T to conventional signalized intersections, using treatment and comparison site data from Florida and South Carolina. The results show that the expected total, fatal and injury, and target crash (rear-end, angle, and sideswipe) frequencies are lower at the continuous green T intersection relative to the conventional signalized intersection (CMFs of 0.958 [95% CI=0.772-1.189], 0.846 [95% CI=0.651-1.099], and 0.920 [95% CI=0.714-1.185], respectively).

Quantifying the safety effects of horizontal curves on two-way, two-lane rural roads.

The objective of this study is to quantify the safety performance of horizontal curves on two-way, two-lane rural roads relative to tangent segments. Past research is limited by small samples sizes, outdated statistical evaluation methods, and unreported standard errors. This study overcomes these drawbacks by using the propensity scores-potential outcomes framework. The impact of adjacent curves on horizontal curve safety is also explored using a cross-sectional regression model of only horizontal curves. The models estimated in the present study used eight years of crash data (2005-2012) obtained from over 10,000 miles of state-owned two-lane rural roads in Pennsylvania. These data included information on roadway geometry (e.g., horizontal curvature, lane width, and shoulder width), traffic volume, roadside hazard rating, and the presence of various low-cost safety countermeasures (e.g., centerline and shoulder rumble strips, curve and intersection warning pavement markings, and aggressive driving pavement dots). Crash prediction is performed by means of mixed effects negative binomial regression using the explanatory variables noted previously, as well as attributes of adjacent horizontal curves. The results indicate that both the presence of a horizontal curve and its degree of curvature must be considered when predicting the frequency of total crashes on horizontal curves. Both are associated with an increase in crash frequency, which is consistent with previous findings in the literature. Mixed effects negative binomial regression models for total crash frequency on horizontal curves indicate that the distance to adjacent curves is not statistically significant. However, the degree of curvature of adjacent curves in close proximity (within 0.75 miles) was found to be statistically significant and negatively correlated with crash frequency on the subject curve. This is logical, as drivers exiting a sharp curve are likely to be driving slower and with more awareness as they approach the next horizontal curve.

A multivariate spatial crash frequency model for identifying sites with promise based on crash types.

Many studies have proposed the use of a systemic approach to identify sites with promise (SWiPs). Proponents of the systemic approach to road safety management suggest that it is more effective in reducing crash frequency than the traditional hot spot approach. The systemic approach aims to identify SWiPs by crash type(s) and, therefore, effectively connects crashes to their corresponding countermeasures. Nevertheless, a major challenge to implementing this approach is the low precision of crash frequency models, which results from the systemic approach considering subsets (crash types) of total crashes leading to higher variability in modeling outcomes. This study responds to the need for more precise statistical output and proposes a multivariate spatial model for simultaneously modeling crash frequencies for different crash types. The multivariate spatial model not only induces a multivariate correlation structure between crash types at the same site, but also spatial correlation among adjacent sites to enhance model precision. This study utilized crash, traffic, and roadway inventory data on rural two-lane highways in Pennsylvania to construct and test the multivariate spatial model. Four models with and without the multivariate and spatial correlations were tested and compared. The results show that the model that considers both multivariate and spatial correlation has the best fit. Moreover, it was found that the multivariate correlation plays a stronger role than the spatial correlation when modeling crash frequencies in terms of different crash types.

Estimating the safety effects of lane widths on urban streets in Nebraska using the propensity scores-potential outcomes framework.

A sufficient understanding of the safety impact of lane widths in urban areas is necessary to produce geometric designs that optimize safety performance for all users. The overarching trend found in the research literature is that as lane widths narrow, crash frequency increases. However, this trend is inconsistent and is the result of multiple cross-sectional studies that have issues related to lack of control for potential confounding variables, unobserved heterogeneity or omitted variable bias, or endogeneity among independent variables, among others. Using ten years of mid-block crash data on urban arterials and collectors from four cities in Nebraska, crash modification factors (CMFs) were estimated for various lane widths and crash types. These CMFs were developed using the propensity scores-potential outcomes methodology. This method reduces many of the issues associated with cross-sectional regression models when estimating the safety effects of infrastructure-related design features. Generalized boosting, a non-parametric modeling technique, was used to estimate the propensity scores. Matching was performed using both Nearest Neighbor and Mahalanobis matching techniques. CMF estimation was done using mixed-effects negative binomial or Poisson regression with the matched data. Lane widths included in the analysis included 9ft, 10ft, 11ft, and 12ft. Some of the estimated CMFs were point estimates while others were functions of traffic volume (i.e., the CMF changed depending on the traffic volume). Roadways with 10ft travel lanes were found to experience the highest crash frequency relative to other lane widths. Meanwhile, roads with 9ft travel lanes were found to experience the lowest relative crash frequency. While this may be due to increased driver caution when traveling on narrow lanes, it is possible that unobserved factors influenced this result. CMFs for target crash types (sideswipe same-direction and sideswipe opposite-direction) were consistent with the values currently used in the Highway Safety Manual (HSM).

Comparison of safety effect estimates obtained from empirical Bayes before-after study, propensity scores-potential outcomes framework, and regression model with cross-sectional data.

A variety of different study designs and analysis methods have been used to evaluate the performance of traffic safety countermeasures. The most common study designs and methods include observational before-after studies using the empirical Bayes method and cross-sectional studies using regression models. The propensity scores-potential outcomes framework has recently been proposed as an alternative traffic safety countermeasure evaluation method to address the challenges associated with selection biases that can be part of cross-sectional studies. Crash modification factors derived from the application of all three methods have not yet been compared. This paper compares the results of retrospective, observational evaluations of a traffic safety countermeasure using both before-after and cross-sectional study designs. The paper describes the strengths and limitations of each method, focusing primarily on how each addresses site selection bias, which is a common issue in observational safety studies. The Safety Edge paving technique, which seeks to mitigate crashes related to roadway departure events, is the countermeasure used in the present study to compare the alternative evaluation methods. The results indicated that all three methods yielded results that were consistent with each other and with previous research. The empirical Bayes results had the smallest standard errors. It is concluded that the propensity scores with potential outcomes framework is a viable alternative analysis method to the empirical Bayes before-after study. It should be considered whenever a before-after study is not possible or practical.