Stroop Test

Most online Stroop tests hand you a made-up score that blends speed and errors into a number with no units. This one gives you the thing the effect actually is.

We take your average time on the trials where the word and the ink disagree. Then we subtract your average time on the trials where they match. Only correct answers count.

That subtraction is what makes the number travel. Both halves of the test share your screen, your keyboard, your eyesight and how tired you are. Subtract one from the other and all of it cancels out. What survives is what it costs you to override a habit, which is what the task was built to expose.

John Ridley Stroop's doctoral work compared two jobs. One was reading colour words printed in the wrong ink. The other was naming plain patches of colour.

The results were lopsided. Naming the ink was slowed badly by the word. Reading the word was barely touched by the ink. If you can read, reading happens whether you want it to or not. Habits like that are cheap to run and expensive to stop.

MacLeod counted hundreds of variations in a 1991 review, and the effect survives nearly all of them. Spoken answers, key presses, other languages, even versions where the clashing word sits beside the patch rather than inside it. It is one of the most repeatable findings in the field. That is exactly why it is worth measuring on yourself rather than reading about.

The effect is largest when you say the colour out loud. The thing fighting you is a written word, and your answer is also a word, so the two collide directly. Pressing a key is a different kind of answer. Key-press versions give smaller and shakier numbers than spoken ones.

A browser cannot time your voice reliably, so this is a key-press version. Its numbers cannot be compared with card-based or voice-recorded clinical Stroop scores.

It also means part of your score is the lookup itself. You have to turn 'blue' into 'the third key', and that step costs real time. What the Stroop effect actually shows covers the limits in more detail than fits here.

This test uses the classic four: red, blue, yellow and green. Red and green are the exact pair that around one man in twelve, and one woman in two hundred, cannot reliably tell apart. No palette that keeps the standard Stroop set can fix that.

Swapping in colours that avoid the clash would make the task fairer. It would also make it a different task, no longer comparable with ninety years of published work. So we kept the standard set, and we are telling you the cost rather than hiding it.

What we can do, we have done. Every button carries the colour name in words next to its patch, so nothing is signalled by colour alone. The task also runs entirely from the keyboard.

If two of the four inks look alike to you, your score will be inflated by mistakes that have nothing to do with attention. That is why we report errors separately. Throw away any run with more than about one wrong answer in ten rather than trying to read it.

A typical key-press Stroop score lands around 75 ms, with most people between 20 and 150. A small number is not proof of superior control. A large one is not a fault. How cautious you are, how well you know the keys, and whether you keep glancing down all move it by tens of milliseconds.

The useful comparison is your own, across days. Sleep and alcohol both move this number.

For a second angle on the same ability, the go/no-go test measures self-control in errors rather than milliseconds. For how much any reaction-based number shifts with age, see average reaction time by age.