Addressing selective reporting bias in meta-analysis of dependent effect sizes: A tutorial in R

Area of Science:

• Psychology
• Biostatistics
• Research Methodology

Background:

• Selective reporting bias occurs when study results are incompletely published based on their magnitude or significance.
• This bias can lead to systematically skewed parameter estimates in conventional meta-analysis.
• Existing methods primarily address univariate meta-analysis, leaving a gap for dependent effect sizes common in psychology.

Purpose of the Study:

• To provide a guide for investigating and correcting selective reporting bias in meta-analyses involving dependent effect sizes.
• To review and demonstrate recently developed statistical methods for bias diagnosis and correction.
• To enhance the reliability of meta-analytic findings in psychology.

Main Methods:

• Review of recently developed statistical techniques for selective reporting bias.
• Demonstration of a regression-based adjustment technique.
• Application of a step-function selection model and a sensitivity analysis approach.
• Implementation in the R statistical environment using example meta-analyses.

Main Results:

• The tutorial illustrates practical application of bias correction methods.
• It highlights the strengths and limitations of each reviewed technique.
• Provides a framework for assessing and mitigating bias in complex meta-analyses.

Conclusions:

• Addressing selective reporting bias is critical for accurate meta-analytic conclusions, especially with dependent effect sizes.
• The discussed methods offer valuable tools for researchers in psychology and related fields.
• Accurate reporting and bias correction improve the validity of synthesized research findings.