Peer-Reviewed Journal Publications

+ Indicates one of Dr. Wood’s Graduate Students

* Indicates a student co-author from Iowa State University

** Indicates a student co-author from Penn State University

*** Indicates a student co-author from South Dakota State University

Abstract: Secondary incidents are considered a major risk in terms of traffic management due to dangerous ramifications, such as reduced capacity, additional traffic delays, and serious injuries. Therefore, it is necessary to examine what causes a secondary incident to occur after a primary incident and prepare countermeasures to reduce the possible damage to human and property from primary and secondary incidents. In Iowa, a safety service patrol program is being implemented on major highway routes to respond to both types of incident efficiently. However, research on when, where, and under what conditions these incidents occur and how the program can deal with incidents must be conducted to determine the major characteristics of primary and secondary incidents and to estimate the program's performance. Consequently, statistical and spatial analyzes were performed on traffic incidents in a 5-year period (2016–2020) in Iowa. A survival analysis confirmed that the program could decrease the probability of secondary incident occurrences, and 99.9% of secondary incidents occurred within 4 h of the primary incident. Additionally, the binary logistic regression analysis of primary incidents affirmed that a longer incident clearance time and a higher severity of incidents could increase the probability of secondary incidents occurrence. Furthermore, a spatial analysis evaluated that the Iowa DOT safety service patrol program adequately covered areas where primary and secondary incidents are focused. This study is expected to be used to develop countermeasures in both incident cases by identifying the characteristics of secondary and primary incidents in Iowa.


Abstract: Secondary incidents are considered a major risk in terms of traffic management due to dangerous ramifications, such as reduced capacity, additional traffic delays, and serious injuries. Therefore, it is necessary to examine what causes a secondary incident to occur after a primary incident and prepare countermeasures to reduce the possible damage to human and property from primary and secondary incidents. In Iowa, a safety service patrol program is being implemented on major highway routes to respond to both types of incident efficiently. However, research on when, where, and under what conditions these incidents occur and how the program can deal with incidents must be conducted to determine the major characteristics of primary and secondary incidents and to estimate the program's performance. Consequently, statistical and spatial analyzes were performed on traffic incidents in a 5-year period (2016–2020) in Iowa. A survival analysis confirmed that the program could decrease the probability of secondary incident occurrences, and 99.9% of secondary incidents occurred within 4 h of the primary incident. Additionally, the binary logistic regression analysis of primary incidents affirmed that a longer incident clearance time and a higher severity of incidents could increase the probability of secondary incidents occurrence. Furthermore, a spatial analysis evaluated that the Iowa DOT safety service patrol program adequately covered areas where primary and secondary incidents are focused. This study is expected to be used to develop countermeasures in both incident cases by identifying the characteristics of secondary and primary incidents in Iowa.


Abstract: This study develops a procedure to rank agencies based on their incident responses using roadway clearance times for crashes. This analysis is not intended to grade agencies but to assist in identifying agencies requiring more training or resources for incident management. Previous NCHRP reports discussed usage of different factors including incident severity, roadway characteristics, number of lanes involved and time of incident separately for estimating the performance. However, it does not tell us how to incorporate all the factors at the same time. Thus, this study aims to account for multiple factors to ensure fair comparisons. This study used 149,174 crashes from Iowa that occurred from 2018 to 2021. A Tobit regression model was used to find the effect of different variables on roadway clearance time. Variables that cannot be controlled directly by agencies such as crash severity, roadway type, weather conditions, lighting conditions, etc., were included in the analysis as it helps to reduce bias in the ranking procedure. Then clearance time of each crash is normalized into a base condition using the regression coefficients. The normalization makes the process more efficient as the effect of uncontrollable factors has already been mitigated. Finally, the agencies were ranked by their average normalized roadway clearance time. This ranking process allows agencies to track their performance of previous crashes, can be used in identifying low performing agencies that could use additional resources and training, and can be used to identify high performing agencies to recognize for their efforts and performance.


Abstract: Crash frequency modeling has been an active research topic in traffic safety, for which various techniques have been proposed that can be loosely classified as either statistical models or machine learning (ML) methods. Statistical models are suitable for drawing inferences and producing relationships that are verifiable by domain experts. However, they generally suffer from low predictive performance due to built-in assumptions about the crash data and adherence to prespecified functional forms. On the other hand, ML methods are data-driven and free from pre-supposed conditions on the dataset, yet they are often not interpretable. In this paper, a combination scheme is proposed to leverage the advantages of both techniques, and it is evaluated using crash data collected from urban highways in the state of Washington. The results show that this combination scheme could significantly improve the predictive performance and model fitness of statistical models without adversely impacting their interpretability. 


Abstract: One critical aspect of bus service quality that influences riders’ attitudes is the availability of seating and/or space to board vehicles. Unfortunately, little attention has been given to short-term passenger occupancy predictions on individual buses. This research examines the use of conventional linear regression models and a machine-learning (random forest) model to predict passenger occupancies on individual buses when they arrive at future stops using data available in real-time from bus operations (e.g., Automatic Passenger Counter (APC) systems) and weather information. Overall, the linear model (LM) and the random forest (RF) model are found to provide close estimates. Three sets of models are developed in this work to model the current and future stop pairs: a next-stop-based model that only models the occupancy at the right next stop and two models that predict the occupancy at any future stop along the bus route (called OD-pair based models). The OD-pair based models are found to predict passenger occupancies more accurately at downstream stops, regardless of whether the LM or RF is used. Examination of the transferability reveals that models can provide reliable estimates of future data when trained with historical information if demand patterns are fairly stable. These models and insights can be used by transit agencies in improving the quality and breadth of information provided to transit system users and even be integrated directly into real-time end-user feeds.


Abstract: The objectives of this study were to compare the adult occupant injury risk on specific body regions in frontal and rear impact and to investigate the effect of those crash conditions and occupant-related factors on the injury risk. Data from the NASS-CDS and Crash Investigation Sampling System were studied for crashes during 2000 to 2019 involving model year 2000 to 2020 motor vehicles, including frontal collisions and rear-end collisions. The injury risk by specific body regions were compared by descriptive statistics, and logistic regression models were developed to examine the effects of various factors on injury risk by specific body regions, controlling for crash type (frontal impact and rear impact), vehicle impact speed, vehicle impact location, vehicle model year, and occupant gender, age, belt use, and seating position. After controlling for the confounding factors, the occupants in frontal impact had higher overall injury risk than in rear impact (at Maximum Abbreviated Injury Scale [MAIS] 3+; odds ratio [OR] = 6.23; 95% confidence interval [CI] [6.06–6.40]), except for lower neck/spine injury risk at MAIS 1+ (OR = 0.47; 95% CI [0.46–0.47]). The impact speed (at MAIS 3+; OR = 1.10; 95% CI [1.10–1.10]) and aging (at MAIS 3+; OR = 1.05; 95% CI [1.05–1.05]) increase overall injury risk, and the unbelted occupants had higher overall injury risk than belted occupants not only in frontal impact (at MAIS 3+; OR = 4.04; 95% CI [3.98–4.10]), but also in rear impact (at MAIS 3+; OR = 28.4; 95% CI [26.4–30.5]). Females had higher overall injury risk than males in frontal impact (at MAIS 3+; OR = 2.01; 95% CI [1.99–2.04]) but not in rear impact (at MAIS 3+; OR = 0.77; 95% CI [0.73–0.81]). Occupants in rear impact had lower injury risk than in frontal impact at MAIS 1+ to MAIS 3+, except for neck/spine at MAIS 1+. The belt restraint was effective not only in frontal impact but also in rear impact. This study provided injury risk references for current vehicles that may provide insight to the potential injury risk of rear-facing occupants in future vehicle configurations.


Abstract: The rear-facing seat has become a potential seating configuration for the future autonomous vehicle, therefore the potential injury risk of a rear-facing occupant merits further investigation not only for adults, but also for children. The objectives of this study were to compare the child occupant injury risk on specific body regions in frontal and rear impact, and to investigate the effect of those crash conditions and occupant related factors on the occupant injury risk. Data from the NASS–CDS and CISS were studied for crashes during 2000–2019 involving model year 2000–2020 motor vehicles, including frontal and rear–end collision. The injury risk by specific body regions were compared by descriptive statistics. Logistic regression models were developed to examine the effects of various factors on injury risk, controlling for crash type (frontal impact and rear impact), vehicle impact speed, vehicle impact location, vehicle model year, and occupant gender, age, restraint use, and seating position. After controlling for the confounding factors, the children age 3–17 had higher injury risk than infant and toddler (age 0–2) (at MAIS 2+, OR 8.71, p < 0.001). The child occupant injury risk was higher in frontal impact than in rear impact (at MAIS 2+, OR 4.96, p < 0.001), especially for children age 3–17 with the exception of the MAIS 1+ neck/spine injury risk. This study provided child injury risk references for current vehicle which may provide insight to the potential injury risk of rear-facing child occupant in future vehicle configurations.


Abstract: Recommender systems attempt to identify and recommend the most preferable item (product-service) to individual users. These systems predict user interest in items based on related items, users, and the interactions between items and users. We aim to build an auto-routine and color scheme recommender system for home-based smart lighting that leverages a wealth of historical data and machine learning methods. We utilize an unsupervised method to recommend a routine for smart lighting. Moreover, by analyzing users’ daily logs, geographical location, temporal and usage information, we understand user preferences and predict their preferred light colors. To do so, users are clustered based on their geographical information and usage distribution. We then build and train a predictive model within each cluster and aggregate the results. Results indicate that models based on similar users increases the prediction accuracy, with and without prior knowledge about user preferences.


Abstract: There has always been a gap between research and engineering practices across almost all disciplines. The nature of these gaps are not identified; However, in order to be able to bridge these gaps a comprehensive understanding is required. In this study, in order to provide research agenda, we examine existing literature and users perception in smart lighting business. We consider two sources of data for this analysis 1-lighting research literature that reflects the research phase and 2-Amazon reviews about two smart lighting products that represent customers’ perceptions reflecting the development phase. We use text mining and analyze both data sources by topic modeling and sentiment analysis. We conducted a probabilistic aspect based sentiment analysis. One of the results of this analysis showed that while the research community is drastically concerned about energy consumption, the end users were excited about color changing ability in smart lighting products.


Abstract: Perception-reaction time (PRT) and deceleration rate are two key components in geometric design of highways and streets. Combined with a design speed, they determine the minimum required stopping sight distance (SSD). Current American Association of Highway Transportation Officials (AASHTO) SSD guidance is based on 90th percentile PRT and 10th percentile deceleration rate values from experiments completed in the mid-1990s. These experiments lacked real-world distractions, and so forth. Thus, the values from these experiments may not be applicable in real-world scenarios. This research evaluated (1) differences in PRTs and deceleration rates between crash and near-crash events and (2) developed predictive models for PRT and deceleration rate that could be used for roadway design. This was accomplished using (1) genetic matching (with Rosenbaum’s sensitivity analysis) and (2) quantile regression. These methods were applied to the Strategic Highway Research Program 2 (SHRP2) Naturalistic Driving Study (NDS) data.

The analysis results indicated that there were differences in PRT and deceleration rates for crash and near-crash events. The specific estimates were that, on average, drivers involved in crash events took 0.487 s longer to react and decelerated at 0.018 g’s (0.58 ft/s^2) slower than drivers in equivalent near-crashes. Prediction models were developed for use in roadway design. These models were used to develop tables comparing existing SSD design criteria with SSD criteria based on the results of the predictive models. These predicted values indicated that minimum design SSD values would increase by 10.5–129.2 ft, dependent on the design speed and SSD model used.


Abstract: 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.


Abstract: Horizontal curves are a contributing factor to the number of observed roadway crashes. Identifying locations and geometric characteristics of horizontal curves plays a crucial role in crash prediction and prevention. However, most states in the USA face a challenge in maintaining detailed and high-quality roadway inventory databases for low-volume rural roads due to the labor-intensive and time-consuming nature of collecting and maintaining the data. This paper proposes a low-cost mobile road inventory system for two-lane horizontal curves based on off-the-shelf smartphones. The proposed system is capable of accurately detecting horizontal curves by exploiting a K-means machine learning technique. Butterworth low-pass filtering is applied to reduce sensor noise. Extended Kalman filtering is adopted to improve the GPS accuracy. Chord method-based radius computation and superelevation estimation are introduced to achieve accurate and robust results despite the low-frequency GPS and noisy sensor signals obtained from smartphones. This study implements this method using an Android-based smartphone and tests 21 horizontal curves in South Dakota. The results demonstrate that the proposed system achieves high curve identification accuracy as well as high accuracy for calculating curve radius and superelevation.


Abstract: The Highway Safety Manual provides multiple methods that can be used to identify sites with promise (SWiPs) for safety improvement. However, most of these methods cannot be used to identify sites with specific problems. Furthermore, given that infrastructure funding is often specified for use related to specific problems/programs, a method for identifying SWiPs related to those programs would be very useful. This research establishes a method for Identifying SWiPs with specific issues. This is accomplished using two safety performance functions (SPFs).This method is applied to identifying SWiPs with geometric design consistency issues. Mixed effects negative binomial regression was used to develop two SPFs using 5 years of crash data and over 8754 km of two-lane rural roadway.The first SPF contained typical roadway elements while the second contained additional geometric design consistency parameters. After empirical Bayes adjustments, sites with promise (SWiPs) were identified. The disparity between SWiPs identified by the two SPFs was evident; 40 unique sites were identified by each model out of the top 220 segments. By comparing sites across the two models, candidate road segments can be identified where a lack design consistency may be contributing to an increase in expected crashes. Practitioners can use this method to more effectively identify roadway segments suffering from reduced safety performance due to geometric design inconsistency, with detailed engineering studies of identified sites required to confirm the initial assessment.


Abstract: When a vehicle departs from the roadway, the characteristics of the roadside influence the probability and severity of a crash. In the American Association of State Highway and Transportation Officials’ (AASHTO) Highway Safety Manual (HSM), the characteristics of the roadside are codified using Roadside Hazard Ratings (RHR). The RHR is a 7-point scale that was developed more than 25 years ago to evaluate the safety performance of cross-sectional roadside design elements. While the HSM provides a function to determine a Crash Modification Factor (CMF) for the RHR, using a rating of 3 as the baseline (CMF = 1.0 for the baseline), no standard errors or confidence intervals are available for these CMFs, so the precision of the CMFs is not well known or understood. Aside from the CMFs found in the HSM, the Federal Highway Administration’s CMF Clearinghouse does not contain any other CMFs for RHR. This study developed CMFs (with associated confidence intervals) for

the 7-point RHR scale that is used in highway safety evaluations using a rating of 3 as a baseline value. Crash and roadway data from Pennsylvania for 21,340 two-lane road segments over an eight-year period (2005-2012), including RHRs ranging from 1 to 7, were used. A causal inference framework utilizing genetic matching in conjunction with random parameters negative binomial regression was used to develop the CMFs. It is expected that the results will be useful for supplementing the HSM RHR CMFs. 


Abstract: 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.


Abstract: Stop-controlled intersections are dangerous locations in which drivers must negotiate conflicts between traffic streams. This study examined driver stopping compliance at minor-street stop-controlled intersections in Qatar. Several variables that may influence driver behavior when approaching a stop sign were considered—including age, gender, driver heritage differences, vehicle type, presence of an approaching vehicle or a pedestrian, peak and nonpeak hours, weekday versus weekend, and residential versus commercial land use. Binary and ordinal logistic regression models were developed to describe driver stopping behavior as functions of these characteristics. Results indicated an alarmingly low compliance rate with minor-street stop signs. Generally, results indicated that male drivers, young drivers, and SUV drivers are less likely to come to a complete stop at these locations. Results also revealed that drivers are more likely to ignore the stop sign when they notice a vehicle or pedestrian approaching and most likely to beat the approaching vehicle or pedestrian through the intersection and reduce intersection-related delay. These findings raise a major safety concern and indicate aggressive driving tendencies. Potential countermeasures include increasing police enforcement, initiating traffic safety campaigns (e.g., targeting the higher risk drivers identified in the results), and improving the visibility of stop signs using different measures, such as larger and additional stop signs, “Stop Ahead” advance traffic control signs, and enhanced pavement markings.


Abstract: Stopping sight distance (SSD) is an important design criterion used in the geometry of highways and streets. Design guidance implies that SSD is used to ensure safety along the roadway. This paper reviews SSD design criteria and develops an updated model to improve consistency between available sight distance and SSD criteria found in geometric design policy. A new variable, the distance from the front of the car to the driver’s eye (Lfront-eye), is used in the updated model. Distributional values for Lfront-eye are determined. A method accounting for lighted (daytime and lighted nighttime) versus unlighted nighttime conditions is also discussed. A probabilistic analysis of vertical curve SSD uses Monte Carlo simulation. The results of this analysis are compared with the SSD model found in current geometric design policy. Possible values for Lfront-eye that can be used in design guidance are proposed. Potential issues that should be investigated in future work are discussed.


Abstract: Transit agencies often provide travelers with point estimates of bus travel times to downstream stops to improve the perceived reliability of bus transit systems. Prediction models that can estimate both point estimates and the level of uncertainty associated with these estimates (e.g., travel time variance) might help to further improve reliability by tempering user expectations. In this paper, accelerated failure time survival models are proposed to provide such simultaneous predictions. Data from a headway-based bus route serving the Pennsylvania State University-University Park campus were used to estimate bus travel times using the proposed survival model and traditional linear regression frameworks for comparison. Overall, the accuracy of point estimates from the two approaches, measured using the root-mean-squared errors (RMSEs) and mean absolute errors (MAEs), was similar. This suggests that both methods predict travel times equally well. However, the survival models were found to more accurately describe the uncertainty associated with the predictions. Furthermore, survival model estimates were found to have smaller uncertainties on average, especially when predicted travel times were small. Tests for transferability over time suggested that the models did not over-fit the dataset and validated the predictive ability of models established with historical data. Overall, the survival model approach appears to be a promising method to predict both expected bus travel times and the uncertainty associated with these travel times.


Abstract: Current engineering practice uses a point-mass model to design horizontal curves on highways and streets. In this model, a maximum side friction factor is used, in combination with the selected design speed and maximum rate of superelevation, to determine the minimum radius of the curve for an alignment. The limiting value for side friction used in design was established in the 1940s and was based on driver comfort thresholds. The lateral friction available at the tire–roadway interface is a measure of friction supply and is dependent on the pavement surface type and condition, vehicle operating speed and deceleration characteristics, vehicle lane position, and tire type. The drivers’ selection of individual operating speeds on a roadway results in a side friction demand when traversing a horizontal curve. The purpose of this paper is threefold. First, key side friction concepts in horizontal curve design are described. This description includes the definitions of and the fundamental principles associated with the application of side friction factors in horizontal curve design policy. Second, the paper provides an analysis of the margin of safety in horizontal curve design policy. This analysis considers various vehicle types, pavement surface types, and operating speed distributions, and makes comparisons between friction supply, demand, and design side friction factors. Third, the paper describes a framework for more effective consideration of the current vehicle fleet, range of pavement conditions, and vehicle speed distribution in horizontal curve design policy.


Abstract: 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.


Abstract: 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).


Abstract: 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 9 ft, 10 ft, 11 ft, and 12 ft. 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 10 ft travel lanes were found to experience the highest crash frequency relative to other lane widths. Meanwhile, roads with 9 ft 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).


Abstract: 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.


Abstract: Criteria for the design of horizontal curves implicitly rely on design speed to produce safe and efficient designs, particularly for horizontal sight line offsets and stopping sight distances. Current design guidance provides a method for calculating minimum horizontal sight line offsets that is accurate and valid only when both driver and object are within the limits of the curve. Other methods are available to estimate minimum horizontal sight line offsets when the driver, object, or both are not within the curve limits. However, design guidance recommends using the calculated value for offsets as a conservative estimate near the ends of curves. In this study, speed prediction models and reliability theory were used to estimate the probability that drivers would not have enough sight distance to see, react to, and stop before reaching an object in the roadway if horizontal sight line offset criteria were applied when the driver or object was outside the limits of a horizontal curve. Six scenarios at the curve approach and inside the curve were analyzed. Reliability estimates (based on minimum horizontal sight line offsets from current minimum design criteria) and stopping sight distance distributions (based on individual driver characteristics) indicated that the probability of drivers not having enough stopping sight distance was much greater on the approach to than inside the horizontal curves. For improvement of design consistency, the use of calculated horizontal sight line offsets beyond the limits of the curve (approach and departure tangents) is suggested to provide extra sight distance to drivers near the curve.


Abstract: Safety (in terms of expected crash frequency and severity) was compared on road segments where design exceptions had been approved and constructed and on similar road segments where no design exceptions had been approved or constructed. Data were collected for design exceptions in Utah from 2001 to 2006. Multiple data sources were used to identify and define road segments with and without design exceptions. Propensity scores were applied to assess similarities between treatment and comparison sites. Ultimately, 34 total nonfreeway segments with design exceptions and 80 nonfreeway segments without design exceptions were used for modeling. The relationship between the presence of design exceptions and crash frequency was explored with a negative binomial regression modeling approach. The relationship between the presence of design exceptions and crash severity was explored in three ways: (a) computation of severity distributions at locations with and without design exceptions, (b) estimation of separate negative binomial regression models by severity level, and (c) estimation of multinomial logit models to predict the severity outcome of a crash. The presence of design exceptions was represented in the regression models by an indicator variable (where 1 5 one design exception or more and 0 5 no design exceptions). Crash data from 2007 through 2010 were used for model estimation. No significant differences were observed in expected crash frequencies and crash severities between nonfreeway road segments with and without design exceptions. This overall finding was consistent with two previous related efforts in Kentucky and Indiana.

Note: This paper won a "Best Paper Award" from the Geometric Design Committee of the Transportation Research Board


Abstract: This paper uses a simultaneous equation modeling approach to explore the relationships between mean speed, standard deviation of speed and work zone design characteristics. Data for model estimation were collected in 17 work zones on four-lane, divided freeways in Pennsylvania and Texas. The three-stage least squares estimator was used, as it is both consistent and efficient in the presence of endogeneity and error covariance. The fixed effects and random effects estimators with instrumental variables were also used to address possible unobserved heterogeneity. Results pointed towards a truly simultaneous structure, where mean speed is determined by exogenous design and traffic control features and the standard deviation of speed and the standard deviation of speed is dependent on design and traffic control features as well as mean speed. A number of work zone design and traffic control features directly influenced both speed parameters. Findings of this research indicate the relationship between speed magnitude, speed dispersion and work zone design and traffic control features is more complex than documented generalizations made in current work zone design and traffic control decision processes.