Inferring Heterogeneous Treatment Effects of Crashes on Highway Traffic - A Doubly Robust Causal Machine Learning Approach

Aug 5, 2026

Upload by Hazel

Highway traffic crashes significantly affect transportation systems and the broader economy, making accurate emergency response planning essential. The impact of crashes on traffic conditions differs across factors and can be distorted by selection bias inherent in observational data. This paper estimates heterogeneous causal effects of crash types on highway speed using a doubly robust causal machine learning framework, combining Neyman-Rubin causal modeling, causal-graph-based adverse variable filtering, and structural causal estimands. Experiments on 4815 Washington State crashes show distance- and duration-varying effects with rigorous testing and sensitivity checks.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages38

Reading Time1 h 36 min

AccessFree to download

Abstract

1. Introduction

Background and motivation

Limitations of conventional regression

Causal inference frameworks

Inferring Heterogeneous Treatment Effects of Crashes on Highway Traffic: A Doubly Robust Causal Machine Learning Approach Shuang Li 1 , Ziyuan Pu 1,2,3*, Zhiyong Cui4 , Seunghyeon Lee5 , Xiucheng Guo 1 , Dong Ngoduy6 Abstract: Highway traffic crashes exert a considerable impact on both transportation systems and the economy. In this context, accurate and dependable emergency responses are crucial for effective traffic management. However, the influence of crashes on traffic status varies across diverse factors and may be biased due to selection bias. Therefore, there arises a necessity to accurately estimate the heterogeneous causal effects of crashes, thereby providing essential insights to facilitate individual-level emergency decision-making. This paper proposes a novel causal machine learning framework to estimate the causal effect of different types of crashes on highway speed. The Neyman-Rubin Causal Model (RCM) is employed to formulate this problem from a causal perspective. The Conditional Shapley Value Index (CSVI) is proposed based on causal graph theory to filter adverse variables, and the Structural Causal Model (SCM) is then adopted to define the statistical estimand for causal effects. The treatment effects are estimated by Doubly Robust Learning (DRL) methods, which combine doubly robust causal inference with classification and regression machine learning models. Experimental results from 4815 crashes on Highway Interstate 5 in Washington State reveal the heterogeneous treatment effects of crashes at varying distances and durations. The rear-end crashes cause more severe congestion and longer durations than other types of crashes, and the sideswipe crashes have the longest delayed impact. Additionally, the findings show that rear-end crashes affect traffic greater at night, while crash to objects has the most significant influence during peak hours. Statistical hypothesis tests, error metrics based on matched “counterfactual outcomes”, and sensitive analyses are employed for assessment, and the results validate the accuracy and effectiveness of our method. Keywords: Highway crashes, Heterogeneous treatment effect, Causal machine learning, Neyman-Rubin Causal Model, Doubly Robust Learning 1 School of Transportation, Southeast University, China 2 Key Laboratory of Transport Industry of ComprehensiveTransportation Theory (Nanjing Modern Multimodal TransportationLaboratory), Ministry of Transport, PRC 3 School of Engineering, Monash University, Malaysia 4 School of Transportation Science and Engineering, Beihang University, China 5 Department of Transportation Engineering, University of Seoul, Korea 6 Department of Civil Engineering, Monash University, Australia 1. Introduction Highway traffic crashes significantly impact highway traffic efficiency, resulting in economic and energy losses. In the United States, the National Highway Traffic Safety Administration estimated that the total cost of car crashes was $340 billion in 2019(Blincore et al., 2023) . Furthermore, highway congestion cost users a total of $45.84 billion in 2019, with more than 18% of this cost related to incidents. To mitigate the adverse effects of congestion resulting from crashes, many highway safetyoriented investigations have primarily concentrated on calculating average effects for emergency planning purposes(Chung and Recker, 2013; Li et al., 2013; Mannering et al., 2016; Chung, 2017; Ren and Xu, 2024) . Nevertheless, considering that the effects of crashes on traffic conditions differ based on factors like pre-crash traffic conditions, location, and time, comprehending heterogeneous effects can significantly enhance the precision of decision-making. Moreover, guaranteeing the credibility of emergency measures necessitates a profound understanding of the causal effects of crashes on traffic conditions, providing crucial support for policymakers. While conventional regression methods are commonly utilized to identify th

FAQs

Why is selection bias a problem when studying crashes and highway speed?

Observational data can create selection bias, leading to biased causal interpretations. Without causal inference, the estimated relationships between crash treatments and speed may understate or overstate true effects.

What causal learning framework does the paper propose?

It proposes a doubly robust causal machine learning approach that uses the Neyman-Rubin causal model to define the causal problem, then employs conditional Shapley value indexing and a structural causal model to specify the estimand.

What major heterogeneous effects are reported in the experiments?

Rear-end crashes produce more severe congestion and longer durations than other crash types, while sideswipe crashes show the longest delayed impact. Effects also vary by time context, such as night and peak hours.

Aug 5, 2026

Upload by Hazel