Cause, effect and regression in road safety: a case study

Area of Science:

• Transportation Safety
• Statistical Modeling
• Road Safety Engineering

Background:

• Evaluating treatment effectiveness in safety is crucial.
• Regression and before-after studies are common methodologies.
• Rail-highway grade crossings serve as a case study for safety interventions.

Purpose of the Study:

• To compare the efficacy of regression analysis versus before-after studies for safety treatment evaluation.
• To assess the reliability and applicability of different study designs in transportation safety.
• To investigate the causal inference capabilities of regression models in safety research.

Main Methods:

• Analysis of published regression studies on 'crossbucks' to 'flashers' replacement at rail-highway grade crossings.
• Examination of before-after study results for the same safety treatment.
• Comparative assessment of methodological strengths and limitations.

Main Results:

• Regression studies produced highly variable results, making causal interpretation difficult.
• Before-after studies showed consistent findings but lacked context-specific applicability.
• Discrepancies in regression models hinder corroboration and validation.

Conclusions:

• Current regression methods struggle to establish reliable cause-and-effect relationships in safety evaluations.
• Before-after studies provide consistent but generalized findings, limiting practical application.
• Crash modification functions from regression offer potential advantages if causal inference is improved.