Analysis of Crash Rates at Freeway Diverge Areas using Bayesian Tobit Modeling Framework

The goal of this study is to evaluate the impact of various risk factors on crash rates at freeway diverge areas. Crash rates data for a three-year period from 367 freeway diverge areas were used for analysis. Four candidate Tobit models were developed and compared under the Bayesian framework: a traditional Tobit model; a random parameters Tobit (RP-Tobit) model; a grouped random parameters Tobit (GRP-Tobit) model; and a random intercept Tobit (RI-Tobit). The results showed that the RP-Tobit model performs best with highest value of Rd2 as well as lowest Mean Absolute Deviance (MAD) and Deviance Information Criteria (DIC), indicating the importance of accounting for unobserved heterogeneity to improve the model fit. Both the GRP-Tobit and the RI-Tobit models provide better performance than the traditional Tobit model. The model results showed that crash rates at freeway diverge areas were positively associated with mainline annual average daily traffic (AADT) and negatively associated with ramp AADT, indicating the different mechanisms of the impact of traffic volume on crash rates at freeway diverge areas. Lane-balanced design and high speed limits at freeway diverge areas have a negative effect on crash rates. The number of lanes on mainline and ramp length have significant heterogeneous effects on crash rates across observations. The RP-Tobit model provides a more comprehensive understanding of the heterogeneous effects of risk factors on crash rates across observations.

Car Drivers’ Experienced Level of Service on Freeways

The Danish Road Directorate sponsored a study to develop methods for quantifying car drivers’ experienced level of service on freeways. The results provide a measure of how well freeways accommodate car travel. To determine how traffic operations, geometric conditions, and other variables affect car drivers’ satisfaction, 188 randomly selected respondents were shown 80 video clips of roadway segments filmed from a driven passenger car. Video clips consisted of high-resolution video filmed through a windshield, side windows including exterior mirrors, and a rear window. Video clips also included a GPS-based speedometer. Respondents rated video clips on a six-point scale ranging from very satisfied to very dissatisfied. The result was 7,497 usable ratings. Four hundred to 450 variables described respondents’ answers to six background questions and the video clips, that is, roadway segment geometries, traffic operations, surroundings, weather, and so forth. Car driver satisfaction models were developed with cumulative logit regression and ordinary generalized linear modeling. The six presented models included three to 10 variables, which related significantly ( p ≤ .05) to satisfaction ratings. These variables were average speed, speed limit, width of hard shoulder, number of entries and other merge areas per mile, number of exits and other diverge areas per mile, flow of long vehicles per lane per hour, direction of sunlight, driver age, type of driver’s license, and driver yearly mileage. Models returned percentage splits of the six levels of satisfaction or average satisfaction. These splits or averages were transformed into a level of service.

Analyzing Crash Frequency by Collision Type at Freeway Diverge Areas

The objective of this study is to analyze the crash frequency by collision type at freeway diverge areas. Three major collision types were identified in this study including rear-end, sideswipe and angle collisions. We compared the crash counts and crash rates by collision type between exit ramp. The data analysis results show that exit ramps with lane-balance design have lower crash frequency and crash rate of rear-end collisions than those not lane-balanced.

Multivariate Poisson-Lognormal Regression for Crash Prediction of Different Types on Freeways Diverge Areas

The primary objective of this study was to develop a multivariate Poisson-lognormal regression model (MVPLN) to evaluate the effects of contributing factors to crash frequency and to identify the correlations among different types. Crash data were collected at 282 exit-ramps on freeways in Florida, US.The Markov chain Monte Carlo (MCMC) method was used to get the solutions of the MVPLN model.The effects of crash-related factors of different types were explored in the MVPLN model considering the correlation structure.It was found thatmainline lane number was the contributing factor for both single-lane and two-lane exit ramp.