Planning sample sizes for before-after accident comparisons.

Area of Science:

• Traffic safety research
• Statistical modeling in transportation engineering

Background:

• Before-and-after safety studies are increasingly favored over cross-sectional studies for establishing crash modification factors.
• The Highway Safety Manual (2010) spurred the adoption of before-and-after study designs.

Purpose of the Study:

• To present a straightforward method for estimating sample sizes required for before-and-after safety studies.
• To achieve a specific level of precision or statistical power when analyzing proportional changes in Poisson rates.

Main Methods:

• Developed a simple statistical method for sample size estimation.
• Assessed the impact of using full Bayesian models on sample size requirements.

Main Results:

• The proposed method provides a way to determine adequate sample sizes for before-and-after safety studies.
• Sample size requirements using full Bayesian models were found to be similar to the proposed simple method.

Conclusions:

• A simple sample size estimation method is effective for before-and-after safety studies.
• Bayesian approaches do not significantly alter sample size needs compared to simpler methods, facilitating practical application in traffic safety analysis.