Meta-analysis relies on standardizing effects to compare across studies. What is the standard metric used?

Meta-analysis relies on standardizing effects to compare across studies. What is the standard metric used?

• A. Averaging p-values across studies
• B. Combining effect sizes across studies to estimate the population effect
• C. Using the mean of study results
• D. Discarding non-significant results from studies

Answer: (B)

Meta-analysis works by expressing each study’s finding as a standardized effect size, a common metric that shows the strength and direction of the effect. This standardization lets us compare across studies that may have used different outcomes or measurement scales. Once we have these effect sizes, we combine them to estimate the overall, or population, effect, typically weighting each study by how precise its estimate is (its variance). This approach preserves information about magnitude and consistency across studies and accounts for study precision and variability. Averaging p-values isn’t appropriate because p-values convey whether an effect is statistically significant but not how large or meaningful the effect is, and they can be influenced by sample size. Simply taking the mean of study results ignores the precision of each study. Discarding non-significant results would bias the synthesis by inflating the estimated effect. The standard metric of meta-analysis is the effect size, used to produce a pooled estimate of the population effect.

Meta-analysis works by expressing each study’s finding as a standardized effect size, a common metric that shows the strength and direction of the effect. This standardization lets us compare across studies that may have used different outcomes or measurement scales. Once we have these effect sizes, we combine them to estimate the overall, or population, effect, typically weighting each study by how precise its estimate is (its variance). This approach preserves information about magnitude and consistency across studies and accounts for study precision and variability.

Averaging p-values isn’t appropriate because p-values convey whether an effect is statistically significant but not how large or meaningful the effect is, and they can be influenced by sample size. Simply taking the mean of study results ignores the precision of each study. Discarding non-significant results would bias the synthesis by inflating the estimated effect. The standard metric of meta-analysis is the effect size, used to produce a pooled estimate of the population effect.