Using beta coefficients to impute missing correlations in meta-analysis research: Reasons for caution.

Area of Science:

• Quantitative Psychology
• Statistical Methods

Background:

• Meta-analysis synthesizes empirical research, often focusing on correlations.
• Primary studies may lack reported correlations, leading to imputation methods.
• Beta estimation procedures (BEPs) are used to impute missing correlations from beta coefficients.

Purpose of the Study:

• To evaluate the accuracy of beta estimation procedures (BEPs) in meta-analysis.
• To assess the impact of BEPs on meta-analytic results.
• To compare the performance of BEPs against using existing correlations.

Main Methods:

• Examined the effect of BEPs on a published meta-analysis.
• Conducted Monte Carlo simulations comparing existing correlations with BEP-imputed data.
• Estimated mean population correlation (ρ̄) and true standard deviation (SDρ) under various conditions.

Main Results:

• BEPs introduced substantial bias in estimating ρ̄ and even larger bias in estimating SDρ.
• Using existing correlations consistently outperformed BEPs.
• BEPs performed worse than using only available correlations in virtually all scenarios.

Conclusions:

• Beta estimation procedures (BEPs) are not recommended for imputing missing correlations in meta-analysis due to significant bias.
• Researchers should prioritize using existing correlations in meta-analyses.
• A return to the standard practice of using only reported correlations is advised for accurate meta-analytic synthesis.