G*Power Style A Priori Sample Size Calculator
Estimate the minimum sample size needed before data collection. This tool supports two common analyses: independent samples t test and correlation.
Common d benchmarks: 0.20 (small), 0.50 (medium), 0.80 (large).
Use 1 for equal group sizes.
Note: This is a fast planning calculator based on standard normal approximations. Desktop G*Power uses exact noncentral distributions and may differ slightly.
Why use a G*Power statistics calculator?
If you run a study with too few participants, you risk missing real effects. If you recruit too many, you waste time, money, and participant effort. A power analysis helps you choose the right sample size before collecting data.
Tools inspired by G*Power are widely used in psychology, education, medicine, business research, and social science. They make your study planning transparent and improve the credibility of your findings.
What this calculator does
This page provides an a priori power analysis, which means you input your expected effect size, significance level, and desired power, and it returns the minimum sample size you should target.
- Independent t test: estimates sample size for comparing two independent means.
- Correlation: estimates sample size for detecting a nonzero Pearson correlation.
How to choose each input
1) Effect size
Your effect size should come from prior literature, pilot data, or a smallest meaningful effect you want to detect.
- Cohen's d (t test): 0.20 small, 0.50 medium, 0.80 large.
- Correlation r: 0.10 small, 0.30 medium, 0.50 large.
When unsure, use a conservative (smaller) effect size. Smaller effects require larger samples.
2) Alpha (α)
Alpha is the Type I error rate, usually set to 0.05. Lower alpha (like 0.01) is stricter and increases required sample size.
3) Power (1 - β)
Power is the probability of detecting an effect if it is truly present. Typical targets are 0.80 or 0.90. Higher power requires more participants.
4) One-tailed vs two-tailed
Two-tailed tests are standard when effects could go in either direction. One-tailed tests can reduce sample size, but should only be used with strong directional justification decided before analysis.
5) Allocation ratio (t test only)
If your groups are unequal, set N2/N1 accordingly. Unequal group sizes generally increase the total sample required for the same power.
Worked examples
Example A: Independent t test
Suppose you expect a medium effect (d = 0.50), set alpha = 0.05, power = 0.80, two-tailed, and equal groups. The calculator will return approximately:
- Group 1: 63 participants
- Group 2: 63 participants
- Total: 126 participants
Example B: Correlation study
If you expect r = 0.30 with alpha = 0.05 and power = 0.80 (two-tailed), the calculator will suggest roughly 84 total participants.
Common mistakes to avoid
- Choosing an unrealistically large effect size just to reduce sample requirements.
- Forgetting attrition and missing data when setting recruitment targets.
- Switching from two-tailed to one-tailed after seeing data.
- Reporting p-values without reporting the power analysis assumptions.
Best-practice checklist for your methods section
- State the software or calculator used for power analysis.
- Report test family, tails, alpha, power target, and effect size.
- Explain where your effect size estimate came from.
- Add planned over-recruitment for dropouts (for example, +10%).
Final note
This calculator is ideal for quick planning and proposal writing. For final preregistration or publication-grade analyses, cross-check with full G*Power software, R packages (such as pwr), or a statisticianāespecially for complex designs like repeated measures, mixed models, or multiple regression with many predictors.