Free data tools for every educator. No PII. No login. No data saved. Built to support you right now.
Small numbers reality check

Can I even trust this number?

A rate built on 12 students behaves nothing like a rate built on 850. One student changing category can move it by 8 points. Enter your rate and the number of students behind it, and we'll show you how much that number can wobble before you act on it. No PII. No login. Everything runs in your browser.

Enter the number you're staring at

No PII
I have
%

Your reality check will appear here

Enter a rate and the number of students behind it. We'll tell you how stable that number really is.

Live, updates as you type

12points of margin of error, plus or minus

Every figure is a student

Every teal figure is a student who met the rate. Every gray figure is a student who didn't. This is the whole group your number is standing on.

The one-student test

8.3 One student changing category moves this number by 8.3 points.

Keep in mind

    How this works

    This tool builds a Wilson 95% score interval around your rate. It's the method statisticians recommend for small groups, because the simple textbook formula falls apart exactly when you need it most: small counts and rates near 0% or 100%. The Wilson interval stays honest in both cases.

    The margin of error is the plain-words version of that interval. It answers one question: if you could rerun this year with a slightly different mix of students, how far could this rate reasonably land from where it is now? A rate of 25% with a margin of error of 3 points is a sturdy number. The same 25% with a margin of error of 24 points is a coin flip wearing a percentage sign.

    The verdict tiers are simple. Under 10 students, we say don't report the number alone, which matches how states handle suppression. A band of 10 points or less is solid enough to act on. Up to 22 points is directional. Wider than that, the number can't carry a decision by itself.

    Sources: Brown, Cai and DasGupta (2001), Interval Estimation for a Binomial Proportion, Statistical Science, and NCES (2017), Best Practices for Determining Subgroup Size.

    A wide margin of error never means a group doesn't matter. It means one number can't carry the story alone. Pair it with trends, context, and what you know about your students.