Are estimates of crash modification factors mis-specified?

Area of Science:

• Transportation Engineering
• Traffic Safety Analysis
• Econometrics

Background:

• Crash modification factors (CMFs) are crucial for evaluating road safety interventions.
• Current CMF estimation often relies on road segment data and geometric features.
• Spatial analysis of crashes incorporates broader network and socio-demographic factors.

Purpose of the Study:

• To investigate omitted variable bias in link-based crash models due to the exclusion of spatial factors.
• To assess the impact of spatial variables on the estimation of CMFs.
• To evaluate the implications for cost-benefit analyses in road safety.

Main Methods:

• Comparison of link-based models with and without spatial factors.
• Statistical analysis to detect omitted variable bias.
• Examination of coefficient sign and magnitude changes.

Main Results:

• Omission of spatial factors results in reduced magnitude of CMF estimates, indicating omitted variable bias.
• Spatial variables show substantial sign changes when integrated into link-based models.
• Significant variability in coefficient estimates was observed.

Conclusions:

• Link-based CMF models may be mis-specified without accounting for spatial factors.
• Omitted variable bias can affect the accuracy of CMFs and subsequent cost-benefit analyses.
• Future road safety research should integrate spatial considerations into CMF estimation.