Free data tools for every educator. No PII. No login. No data saved. Built to support you right now.
Making sense of data

Small Numbers Reality Check

How much should you trust a rate built on a small group?

Your number

Example: 25% of a group of 12. Change any number to use your own.
I have
%

students

Try:
Too shaky

likely range (95%)
points the rate moves when one student changes

Next step

How this works

The likely range is a Wilson 95% score interval: the range that most likely holds the underlying rate, the one this group would show over many similar years. It is the method statisticians recommend for small groups and for rates near 0% or 100%, where the textbook plus-or-minus formula breaks down (Brown, Cai and DasGupta, 2001). Near 0% and 100% the range is lopsided, so the tool shows both ends instead of one plus-or-minus number.

  • Under 10 students: too few to call. Look at the students, not the rate.
  • Range 10 points wide or less: solid enough to plan on.
  • Up to 22 points wide: use with care, directional only.
  • Wider than that: too shaky to compare groups or judge a program on its own.

A percentage is turned into a whole number of students first (25% of 7 becomes 2 of 7, which is 28.6%), because a rate can only be built from whole students.

The caution is about comparing and judging. For a teacher looking at their own students, the count is complete, and acting on those students is always right. Several shaky years moving the same way tell you more than one sharp one.