One 45-minute class. No account, no install, and it runs on paper if the devices fail.
Most Stroop lessons stop at the demonstration: the words are hard to read, everyone laughs, the effect is named, class ends. This one goes one step further, to the step where the actual science lives — students measure the effect, discover that their own number is unreliable, and find the effect anyway once the class pools its data.
That last move is the point of the lesson. It is also the honest one: a single Stroop run is noisy, and pretending otherwise teaches students to trust a number they should not trust.
Open the Stroop test once yourself to see what students will see. Nothing to install and no account to create. Use the Standard round for this lesson — 16 trials, about two minutes. It is the plain version: no response deadline, so the only thing varying between the two conditions is the conflict itself.
The free allowance is 5 rounds per test per day, and it is counted per browser rather than per person. With one device per student this never comes up. On a shared lab machine the students after the first 5 will hit the cap — use phones, or run Path B.
Browser timing is not laboratory timing. Display latency and input lag add tens of milliseconds that vary by machine. The interference effect is large enough to survive that; individual response times from two different devices are not comparable, which is a fact this lesson uses rather than hides.
Students do not need an account and nothing needs to be installed. Signing in exists only to save scores; a student who never signs in can still run every test.
This is closer to what Stroop actually did in 1935: his participants read printed cards aloud while he timed the whole list. Build two cards in any word processor — about five minutes of prep.
Students work in pairs with a phone stopwatch. One names all the ink colors aloud while the partner times the whole card, then they swap. Print one card pair per two students.
Materials. Devices or the two printed cards; something to pool numbers on (a whiteboard or a shared spreadsheet); the exit-ticket prompt.
Show both conditions side by side without running them. Ask every student to write down two things: which condition will be slower, and by how much — a number, in milliseconds on the online path or seconds on paper.
Do not skip the second one. Almost everyone predicts the direction correctly, so direction alone is not interesting. The size is what nobody knows, and having written a number down is what makes the rest of the class land.
Online: each student plays one free round. The task is to name the ink color, not the word. The result screen reports mean response time for matching trials and for conflicting trials separately, which is the raw material for the next step.
On paper: partner-timed, Card A then Card B, whole-card time in seconds, reading across the rows.
The interference effect is the conflicting condition minus the matching condition. Everyone writes down their own number — milliseconds online, seconds on paper — and writes their prediction from step 1 next to it.
Ask for a show of hands: whose measured number was within 50 ms, or one second, of what they predicted? Usually very few. Leave that hanging; the discussion explains it.
Every student's interference effect goes on the board. Then, as a class, find the median — median rather than mean, because one student who sneezed mid-round can drag a mean a long way, and that is itself worth one sentence.
Then count how many students got a number that is zero or negative: no interference, or conflicting trials somehow faster. There will almost always be some. This is the moment the lesson is built around, so do not treat it as a failed measurement.
Work through these in order. The first is the textbook content; the rest is the part that transfers to any measurement a student will ever make.
Collect it on the way out. It is the one place a student has to reason rather than recall.
A friend runs the Stroop test once, gets a negative interference effect, and concludes the Stroop effect is not real. In two or three sentences: what is wrong with that reasoning, and what should they do instead?
| Developing | Proficient | Advanced | |
|---|---|---|---|
| Explains the effect | States that conflicting trials are slower. | Explains it as automatic reading competing with deliberate color naming. | Also explains why practice, rather than difficulty, is what makes reading automatic. |
| Uses the measurement | Computes the difference score. | Computes it and compares it against their own written prediction. | Also states what would have to change for one person's score to be trustworthy. |
| Reasons from evidence | Notices that individual results vary. | Explains that a single run is too noisy to draw a conclusion from. | Explains why pooling reduces the noise while leaving the effect in place. |
The two science practices are the closer match: this lesson is mostly an exercise in reading a small data set honestly and arguing from it.
Standards wording was checked against the official documents on 2026-08-06.
Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18(6), 643–662.
The original used printed cards read aloud and timed as a whole list. The trial-by-trial keyed version taught today came later.
Answers are four keys rather than speech, so what is measured is manual interference, not vocal. The interference effect itself survives that change; the absolute times do not transfer.
This reference was checked against the published record on 2026-08-05. Every other task on the site is sourced the same way on the instructors’ page, together with exactly what a student session records.
The online test is a practice and self-tracking exercise. It is not a validated instrument, it is not normed, and it cannot screen for or indicate any condition. Scores are internal to this site and cannot be compared with published figures, or with another student on another device.
There is no trial-level data export. If you need per-trial output, counterbalancing, or anything you intend to analyze properly, use a platform built for running studies — PsyToolkit, jsPsych and Inquisit all exist for that and this does not replace them. What this is good for is a class that can measure something real in two minutes without an account.
One thing worth raising with a class that asks about scores: the device is part of the measurement. That confound, and why a single number cannot separate several different abilities, is set out on the processing speed page.
This lesson plan is licensed CC BY 4.0. You may copy it, print it, put it in a course pack, translate it, cut it to twenty minutes, or rewrite it for a different age group — including commercially — as long as you credit it.
“Stroop Effect Lesson Plan” by Focani, licensed under CC BY 4.0. https://focani.com/lesson-plans/stroop-effect
What the license covers is the text of this lesson plan. The site itself, the tests and the other pages are not covered by it — those you are welcome to link to freely, with no permission and no attribution needed, which is a different thing from reusing this text.