The experimental design of postmortem studies: the effect size and statistical power

Area of Science:

• Biostatistics
• Forensic Science
• Medical Research

Background:

• Statistical power is crucial for reliable research findings.
• Postmortem studies frequently suffer from insufficient statistical power, potentially leading to erroneous conclusions.
• Understanding effect size is key to performing accurate power analyses.

Purpose of the Study:

• To demonstrate the low statistical power commonly observed in postmortem studies.
• To estimate the effect size specific to postmortem research.
• To provide a basis for power analysis and sample size determination in this field.

Main Methods:

• Utilized GPower software for statistical calculations.
• Set significance level (α) at 0.05 and desired statistical power (1-β) at 0.80.
• Performed calculations for both two-group (Student's t-distribution) and multiple-group (one-way ANOVA) scenarios.

Main Results:

• An average effect size of 0.46 was determined from the analyzed postmortem studies (n=22; SD=0.30).
• Subsequent analysis of a separate group of postmortem studies (n=5) indicated poor average statistical power (1-β < 0.80).

Conclusions:

• Postmortem studies exhibit a significant probability of Type II errors due to low statistical power.
• Implementing power analyses using the estimated effect size can enhance the statistical power of future postmortem research.
• Accurate sample size determination through power analysis is essential for robust postmortem study design.