TY - JOUR
T1 - Crash modification factors of rumble strips on horizontal curves of two-lane rural roads
T2 - A propensity scores potential outcomes approach
AU - Ahmed, Tanveer
AU - Mahmud, Asif
AU - Gayah, Vikash V.
N1 - Publisher Copyright:
© 2023 Elsevier Ltd
PY - 2024/1
Y1 - 2024/1
N2 - Horizontal curves are known to be more crash-prone than tangent sections particularly with respect to roadway departure crashes. Rumble strips are an effective countermeasure to mitigate various types of roadway departure crashes. While existing studies on the safety effectiveness of rumble strips have primarily used before-after study designs or cross-sectional methods for crash modification factor (CMF) estimation, these methods often suffer from imbalanced datasets and larger standard errors, especially when the sample size is small. To address this, this study applies the propensity score potential outcome (PSPO) framework to estimate CMFs for centerline rumble strips, shoulder rumble strips, and their combined application on horizontal curves. In addition to contributing to the development of CMFs by crash severity, this study also examines the effects of rumble strips on collision types, highlighting their impact on vehicle maneuvering and collision characteristics. The analysis is conducted on horizontal curves on two-lane rural roads in Pennsylvania, utilizing crash data from 2017 to 2021. The PSPO method effectively reduces bias between sites with and without rumble strips, and the resulting statistical models align with engineering judgment. The findings indicate that centerline rumble strips reduce opposite direction sideswipe and head-on crashes but increase run off the road and hit fixed object crashes. Shoulder rumble strips, either alone or in combination with centerline rumble strips, decrease crash frequencies for most types except opposite direction sideswipe and head-on crashes. However, shoulder rumble strips alone are more effective at reducing crash frequencies on horizontal curves than when combined with centerline rumble strips.
AB - Horizontal curves are known to be more crash-prone than tangent sections particularly with respect to roadway departure crashes. Rumble strips are an effective countermeasure to mitigate various types of roadway departure crashes. While existing studies on the safety effectiveness of rumble strips have primarily used before-after study designs or cross-sectional methods for crash modification factor (CMF) estimation, these methods often suffer from imbalanced datasets and larger standard errors, especially when the sample size is small. To address this, this study applies the propensity score potential outcome (PSPO) framework to estimate CMFs for centerline rumble strips, shoulder rumble strips, and their combined application on horizontal curves. In addition to contributing to the development of CMFs by crash severity, this study also examines the effects of rumble strips on collision types, highlighting their impact on vehicle maneuvering and collision characteristics. The analysis is conducted on horizontal curves on two-lane rural roads in Pennsylvania, utilizing crash data from 2017 to 2021. The PSPO method effectively reduces bias between sites with and without rumble strips, and the resulting statistical models align with engineering judgment. The findings indicate that centerline rumble strips reduce opposite direction sideswipe and head-on crashes but increase run off the road and hit fixed object crashes. Shoulder rumble strips, either alone or in combination with centerline rumble strips, decrease crash frequencies for most types except opposite direction sideswipe and head-on crashes. However, shoulder rumble strips alone are more effective at reducing crash frequencies on horizontal curves than when combined with centerline rumble strips.
UR - http://www.scopus.com/inward/record.url?scp=85176279519&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85176279519&partnerID=8YFLogxK
U2 - 10.1016/j.aap.2023.107371
DO - 10.1016/j.aap.2023.107371
M3 - Article
C2 - 37948833
AN - SCOPUS:85176279519
SN - 0001-4575
VL - 194
JO - Accident Analysis and Prevention
JF - Accident Analysis and Prevention
M1 - 107371
ER -