Meta-analysis and partial correlation coefficients: A matter of weights

Area of Science:

• Statistics
• Biostatistics
• Meta-analysis

Background:

• A recent study recommended using a suboptimal estimator for partial correlation coefficients (PCCs) standard error in meta-analysis weighting.
• This recommendation was based on Monte Carlo simulations.

Approach:

• This study re-evaluates the simulation framework used in the prior research.
• It explores alternative estimators for PCC standard error beyond the one recommended.
• The performance of different estimators was compared within a simulation environment.

Key Points:

• The previously recommended suboptimal estimator for PCC standard error is not necessarily the best performing.
• Other estimators demonstrated superior performance in the simulations conducted.
• The findings challenge the universality of the prior recommendation.

Conclusions:

• The current evidence is insufficient to establish definitive best practices for meta-analyses using PCCs.
• Further research is required to identify optimal estimators for PCC standard error in various meta-analytic contexts.
• A cautious approach is advised for meta-analysts regarding weighting procedures with PCCs.