A Monte-Carlo simulation analysis for evaluating the severity distribution functions (SDFs) calibration methodology...

Area of Science:

• Highway safety engineering
• Traffic accident analysis
• Statistical modeling

Background:

• Severity distribution functions (SDFs) are crucial for highway safety evaluations but developing new models is resource-intensive.
• The Highway Safety Manual (HSM) provides documented SDF models for various facilities, including freeways and ramps.
• Existing HSM SDF models require local calibration due to data limitations from selected states.

Purpose of the Study:

• To validate the Highway Safety Manual's proposed methodology for calibrating SDFs using a scalar calibration factor.
• To investigate the bias between true and estimated calibration factors under various conditions.
• To establish reliable sample-size guidelines for calibrating SDF models.

Main Methods:

• Extensive simulation analysis was conducted to examine the bias in calibration factors.
• Simulation studies were performed to determine optimal calibration sample sizes.
• Sample-size guidelines were developed based on the average coefficient of variation (CV) of crash severities.

Main Results:

• A significant bias was observed between true and estimated calibration factors, increasing as the true factor deviates from 1.
• The required sample size for calibration increases with higher average coefficients of variation (CV) for KAB and C crashes.
• The study provides data-driven guidelines for selecting appropriate sample sizes for SDF calibration.

Conclusions:

• The HSM's calibration methodology needs further validation, as bias increases with deviation from a calibration factor of 1.
• The coefficient of variation of crash severities is a key factor in determining the necessary sample size for accurate SDF calibration.
• Proposed sample-size guidelines can improve the reliability and efficiency of applying HSM SDF models to new jurisdictions.