A Smooth Bootstrap Procedure towards Deriving Confidence Intervals for the Relative Risk

Area of Science:

• Biostatistics
• Statistical Inference
• Data Analysis

Background:

• Bootstrap methods are frequently used for confidence intervals of relative risk.
• Existing methods may have limitations, particularly with unbalanced sample sizes.

Purpose of the Study:

• To introduce a novel smooth bootstrap procedure for relative risk confidence intervals.
• To evaluate the performance of the new method against existing techniques.

Main Methods:

• Developed a smooth bootstrap procedure using a continuous quantile function.
• Generated pseudo-samples to derive confidence intervals.
• Conducted simulation studies under various settings.

Main Results:

• The proposed smooth bootstrap method demonstrated superior or equal performance compared to asymptotic and existing bootstrap methods.
• The method showed particular advantages for heavily unbalanced data.
• Performance was assessed based on coverage probability and statistical power.

Conclusions:

• The new smooth bootstrap procedure is a robust and effective tool for estimating relative risk confidence intervals.
• It offers significant advantages, especially in scenarios with substantial differences in sample sizes between groups.
• The method's applicability was demonstrated on real-world datasets.