← Back to the tools

What the average hides

Every dashboard you have ever been shown reduces a room full of students to a single number. Sometimes that number is the story. Often it is the thing standing between you and the story. Here are four datasets where the summary looks settled and the data says otherwise. Look at each one before you scroll.

What to do instead

None of this needs new software. It requires looking at the data you already have before you average it.

Plot every pointBefore any mean, look at the distribution. It takes seconds and it is where every one of these stories lived.

Ask who movedAn average change hides who improved, who stayed put, and who went backwards. Those are three different groups needing three different responses.

Check the edgesPiles at the top or bottom of a scale mean your measure has run out of room. No intervention will show up if the ceiling is in the way.

Report the spreadA mean without a spread is half a finding. Give people the range and the shape, not just the midpoint.

Try it on your own numbers

Paste a column of numbers, from a survey, a gradebook, an attendance export, and see every point plotted, with the things worth checking flagged in plain English.

Runs entirely in your browser. Nothing is uploaded, stored, or sent anywhere.

The four cases use simulated data built to make each pattern visible. The patterns themselves are ordinary, and you will meet all four in real student data. Built by Brooklyn Corbett · brooklyncorbett.com