A comparison of Bayesian synthesis approaches for studies comparing two means: A tutorial.

Area of Science:

• Statistical methodology
• Biostatistics
• Quantitative synthesis

Background:

• Synthesizing research findings is crucial for statistical conclusions and effect size estimation.
• Meta-analysis is a common but not the only approach for study synthesis.
• Alternative Bayesian synthesis methods are less discussed but offer valuable alternatives.

Purpose of the Study:

• To illustrate and compare multiple Bayesian synthesis approaches.
• To guide researchers on selecting appropriate methods for combining study data.
• To provide practical implementation details using real data and R code.

Main Methods:

• Comparison of Bayesian meta-analysis, integrative data analyses, and data fusion techniques.
• Application of fixed-, random-, and mixed-effects models.
• Utilizing real-world data for comparative analysis and R code demonstration.

Main Results:

• Detailed comparison of the strengths and limitations of each Bayesian synthesis approach.
• Demonstration of how to apply these methods to combine independent or matched group mean data.
• Provision of R code for practical implementation of all discussed methods and models.

Conclusions:

• Bayesian synthesis offers a flexible framework for combining study results beyond traditional meta-analysis.
• The choice of method depends on the specific research question and data structure.
• Recommendations are provided to aid researchers in future study synthesis efforts.