Flexible models for accidents on Chilean roads

Area of Science:

• Road safety research
• Traffic engineering
• Transportation science

Background:

• Road accidents are a major public health concern.
• Understanding factors influencing accident rates is crucial for mitigation.
• Previous studies often used linear models for traffic flow and road geometry.

Purpose of the Study:

• To analyze the combined effects of traffic flow and road geometry on accident occurrence.
• To compare linear models with a flexible quadratic form for accident prediction.
• To investigate the impact of road geometry on the relationship between traffic flow and accidents.

Main Methods:

• Time-series analysis of cross-sectional data from Chilean paved roads.
• Development and comparison of linear and flexible quadratic regression models.
• Analysis of total accidents and accidents involving pedestrians (AIP).

Main Results:

• Second-order effects of geometric characteristics are highly significant.
• Omitting geometric effects introduces downward bias in the estimated effect of traffic flow.
• Flexible quadratic forms provide a more accurate representation of accident determinants.

Conclusions:

• Road geometry plays a critical role in accident causation, interacting with traffic flow.
• Linear models may underestimate the impact of traffic flow due to unmodeled geometric effects.
• Flexible functional forms are recommended for robust analysis of road safety factors.