Inferences about the between-study variance in meta-analysis with normally distributed outcomes.

Area of Science:

• Biostatistics
• Medical Statistics
• Quantitative Science

Background:

• Meta-analysis is crucial for synthesizing evidence from multiple studies.
• Accurate estimation of between-study variance is essential for reliable meta-analysis results.
• Existing methods for between-study variance estimation may have limitations.

Purpose of the Study:

• To develop a new confidence interval estimation method for between-study variance.
• To utilize the concepts of generalized variables for this estimation.
• To facilitate hypothesis testing through P-value generation.

Main Methods:

• The proposed method is based on generalized variables.
• Confidence intervals for between-study variance are derived.
• A simulation study was conducted to evaluate performance.
• The approach is applicable to normally distributed responses.

Main Results:

• Simulation results indicate satisfactory coverage probabilities for the proposed confidence intervals.
• The method provides a straightforward way to obtain P-values for hypothesis testing.
• The approach is effective for normally distributed data in meta-analysis.

Conclusions:

• The novel approach offers a robust method for confidence interval estimation of between-study variance.
• It is particularly suitable for meta-analyses of controlled clinical trials and epidemiological studies.
• The generalized variables approach enhances inference in meta-analysis.