Sample size calculator
How many subjects you need, with the justification written out and ready to paste into your methods section.
Sample size is the part of the methods section that attracts most rejections and gets documented worst. Not because the arithmetic is hard, but because it has to be explained: what power, what effect size, where that effect came from and why.
This calculator does the arithmetic and, more importantly, writes the paragraph. The number alone does not convince a reviewer; the number with its justification does.
Be the first to use it
It is being built. Leave your email and we will tell you the day it goes live. Nothing else: no newsletter, no promotions.
Used only to notify you about this tool. Ask us to delete it whenever you like.
What it does
The designs people actually use
Two-means comparison, two proportions, correlation, one-way ANOVA, chi-square, and single-group studies estimating a proportion to a given precision.
Methods paragraph, written
Returns the full text with test, power, alpha, effect size and where it came from, in the form reviewers expect.
Power curve
Shows what happens if you recruit fewer subjects than planned, which is what ends up happening. Useful for knowing your margin in advance.
Attrition adjustment
Add the dropout rate you expect and get the number to recruit, not just the number to analyse.
A real example
What you enter
- Design:
- Two independent means
- Effect size (Cohen's d):
- 0.5 (medium)
- Power:
- 80%
- Alpha:
- 0.05, two-tailed
- Expected attrition:
- 15%
What you get
- Per group
- 64 subjects
- Total to analyse
- 128 subjects
- Total to recruit (15% attrition)
- 152 subjects
- Paragraph for the methods section
- Sample size was calculated to detect a difference in means with a medium effect size (Cohen's d = 0.5), assuming 80% statistical power and a two-tailed significance level of 0.05. The calculation, performed for an independent-samples t test, requires 64 subjects per group (128 in total). Anticipating a 15% attrition rate, recruitment was planned for 152 subjects.
The paragraph adapts to the test and the values you entered. It is not a template with blanks.
Why another calculator
There are many, and several are good. None writes the justification, which is exactly the part reviewers send back. And almost all of them ask for the effect size without explaining where to get it, which is where everyone gets stuck.
Frequently asked questions
Do I need an account?
No. It is a calculator: it consumes nothing and needs no sign-up.
Where do I get the effect size?
From a similar previous study, from a pilot, or from the smallest difference you consider clinically relevant. The tool walks you through all three and records which one you used in the written paragraph.
Does it replace G*Power?
Not for complex designs. It covers the common cases and adds what G*Power does not: writing the justification.
When will it be ready?
It is being built. Leave your email and we will tell you the moment it works.
In the meantime
This tool is available today and solves something adjacent:
Manuscript EvaluationBe the first to use it
It is being built. Leave your email and we will tell you the day it goes live. Nothing else: no newsletter, no promotions.
Used only to notify you about this tool. Ask us to delete it whenever you like.