Sample Size Determination
You'll learn how to calculate the minimum sample size needed for a study using your desired statistical power and expected effect size.
The Core Idea
Imagine planning a medical trial before you start: you need to know how many participants to recruit. Too few, and you might miss a real effect (high Type II error). Too many, and you waste resources. Sample size determination uses the concepts you've learned—power, effect size, and significance level—to find the "Goldilocks" sample size that's just right.
The key relationship is:
- Larger effect sizes → need fewer participants to detect
- Higher desired power → need more participants
- Stricter significance level (smaller α) → need more participants
The Formula Approach
For a one-sample z-test (σ known), the approximate formula is:
n ≈ ((z_α/2 + z_β) × σ / δ)²
Where:
n= required sample sizez_α/2= critical value for your significance level (e.g., 1.96 for α=0.05, two-tailed)z_β= critical value for your desired power (e.g., 0.84 for power=0.80)σ= population standard deviationδ= minimum effect size you want to detect (the difference from null hypothesis)
Practical Example
Suppose you want to detect a mean difference of 5 units with σ=10, α=0.05, and power=0.80:
n ≈ ((1.96 + 0.84) × 10 / 5)² = (2.8 × 2)² = 31.36 → need at least 32 subjects
Different tests (t-tests, proportions, ANOVA) have variations of this formula, but the principle remains: you're balancing your tolerance for errors against practical constraints.
Key Takeaway: Sample size determination converts your study goals (desired power and detectable effect size) into a concrete number of observations needed, ensuring your experiment has a reasonable chance of detecting real effects while controlling error rates.