Do speed cameras reduce road traffic collisions?

Area of Science:

• Transportation Safety
• Causal Inference
• Statistical Modeling

Background:

• Previous studies on speed camera effectiveness show varied results.
• Existing methods often overlook causality and confounding factors.
• Outcome regression and simple before-after comparisons are common but limited.

Purpose of the Study:

• To quantify the causal effect of speed cameras on road traffic collisions.
• To address limitations in previous empirical work by employing advanced causal inference.
• To introduce a novel approximate Bayesian doubly-robust (DR) framework for evaluating transport safety interventions.

Main Methods:

• Utilized an approximate Bayesian doubly-robust (DR) causal inference estimation method.
• Combined propensity score (PS) and outcome regression (OR) models for a robust Average Treatment Effect (ATE) estimator.
• Derived posterior predictive distributions for the ATE of speed cameras.

Main Results:

• Identified significant reductions in road traffic collisions at speed camera sites.
• Estimated a mean Average Treatment Effect (ATE) of -15% for collisions.
• Demonstrated the effectiveness of speed cameras in enhancing road safety.

Conclusions:

• The approximate Bayesian DR approach provides a reliable method for evaluating transport safety interventions.
• Speed cameras are an effective tool for reducing road traffic collisions.
• This methodology offers a promising avenue for future transport safety research.