Statistical Inferences Using Effect Sizes in Human Endothelial Function Research.

Area of Science:

• Cardiovascular Research
• Physiology
• Biostatistics

Background:

• Effect size statistics are crucial for quantifying changes in endothelial function research.
• Cohen's guidelines (small d=0.2, medium d=0.5, large d=0.8) are widely used but may not reflect specific research fields.
• Previous analyses show effect size distributions vary across disciplines.

Purpose of the Study:

• To determine the actual distribution of effect sizes in human endothelial function research.
• To establish specific benchmarks for small, medium, and large effects in this field.
• To provide guidance for future research and power calculations in endothelial function studies.

Main Methods:

• Extracted 752 effect sizes (standardized mean differences) from meta-analyses on endothelial function.
• Constructed a frequency distribution to analyze effect size data.
• Utilized the 25th, 50th, and 75th percentiles to define small, medium, and large effects.

Main Results:

• Established new benchmarks for endothelial function research: small effect (d=0.28), medium effect (d=0.69), and large effect (d=1.21).
• These values represent the 25th, 50th, and 75th percentiles of the observed effect sizes.
• The findings indicate Cohen's general guidelines may underestimate effect magnitudes in this specific area.

Conclusions:

• Direct application of Cohen's guidelines would likely underestimate effect magnitudes in human endothelial function research.
• The derived benchmarks are valuable for a priori power analyses and contextualizing findings.
• Encourages consistent reporting of effect sizes to improve the reliability and comparability of endothelial function studies.