A Bayesian aggregate meta-analytic evaluation approach

Area of Science:

• Statistics
• Biostatistics
• Meta-analysis

Background:

• Traditional meta-analysis relies on average effect sizes.
• Limitations exist in sensitivity and consistency of traditional methods.
• Need for robust methods to combine evaluation results, especially without primary data.

Purpose of the Study:

• To modify and evaluate the Bayesian inferential process for aggregate meta-analytic evaluation.
• To compare the Bayesian approach with the traditional average effect size meta-analytic approach.
• To assess the statistical power and consistency of the Bayesian method in meta-analysis.

Main Methods:

• Modification of the Bayesian inferential process for aggregate meta-analysis.
• Comparative analysis of Bayesian and traditional average effect size meta-analytic approaches.
• Evaluation of descriptive and inferential statistics derivation within the Bayesian framework.

Main Results:

• The Bayesian approach demonstrated higher sensitivity to inter-study differences compared to traditional methods.
• The Bayesian approach yielded more consistent methods for deriving descriptive and inferential statistics.
• The Bayesian approach proved statistically more powerful due to its ability to incorporate all available information.

Conclusions:

• The Bayesian inferential process is a superior approach for aggregate meta-analysis.
• Recommended for combining evaluation results when primary data are unavailable.
• Particularly suitable for meta-analyses involving comparisons of two independent samples.