g power calculator

What is a G*Power calculator?

A G*Power calculator helps you plan statistical tests before collecting data. In research design, the goal is usually to avoid two common mistakes:

• Using too few participants and missing real effects (low statistical power).
• Using more participants than needed, which wastes time and resources.

This page focuses on the same type of planning many people do in G*Power software: the independent two-sample t-test for differences in means between two groups.

What this calculator can solve

1) Required sample size

Given alpha, desired power, expected effect size, and group ratio, the calculator estimates how many participants you need per group.

2) Achieved power

If you already know your sample sizes and expected effect size, this mode estimates your power.

3) Minimum detectable effect size

If sample sizes are fixed, this mode tells you the smallest standardized mean difference (Cohen's d) you can reliably detect at your selected alpha and power.

How the math works (quick overview)

This calculator uses a standard normal approximation for two-group testing. For planning, this is often close to G*Power and useful for quick decisions.

• Effect size: Cohen's d = (mean1 - mean2) / pooled SD
• Power target: commonly 0.80 or 0.90
• Alpha: commonly 0.05

For formal preregistration or high-stakes studies, always verify assumptions (variance equality, attrition, multiple testing, clustering, and distribution shape) and run sensitivity checks.

Choosing a realistic effect size

Effect size is usually the hardest input. A few practical options:

• Use results from high-quality prior studies or meta-analyses.
• Use the smallest effect that would matter in practice (minimum clinically or practically important difference).
• Run multiple scenarios (optimistic, realistic, conservative) and compare required sample sizes.

If prior evidence is noisy, avoid planning around large effects. Conservative assumptions reduce the risk of underpowered studies.

Example workflow

Scenario

You expect a medium effect (d = 0.50), want 80% power, alpha = 0.05, and two-tailed testing with equal group sizes.

Using the calculator

• Select Required sample size.
• Set effect size to 0.50.
• Set desired power to 0.80.
• Set ratio to 1.

You will get the required participants in each group and total N. Then add a cushion for expected dropouts (for example, 10% to 20%).

Frequently asked questions

Is this exactly the same as desktop G*Power?

Not exactly. This tool uses a normal approximation for speed and clarity. It is excellent for planning and education, but exact values can differ slightly from noncentral t implementations.

One-tailed or two-tailed?

Use two-tailed unless you have a strong, pre-specified directional hypothesis and a defensible reason to ignore effects in the opposite direction.

What is a good power target?

0.80 is common, while 0.90 is better when feasible, especially in costly or high-impact studies.

Bottom line

A good G*Power calculator makes study planning faster and more transparent. Use it early, test multiple assumptions, and document your choices. Thoughtful power analysis dramatically improves research quality.