The Stroop effect test

The original study timed whole cards of words with a stopwatch; the modern one timed every single answer on a keyboard. The pattern is the same: conflicting words slow you down a lot, conflicting ink hardly at all.

John Ridley Stroop published his experiments in 1935. In one, college students read colour words aloud, once printed in black and once in ink of a different colour. The coloured ink barely mattered: reading 100 words took only 2.3 seconds longer, or 5.6%^[1].

In the other, 100 students named the ink colours instead. Naming the colours of plain squares took 63.3 seconds on average for a card of 100. Naming the ink of words that spelled other colours took 110.3 seconds: an increase of 47 seconds, or 74.3%^[1]. The word got in the way of the colour, but the colour did not get in the way of the word.

For a practised adult, reading a short word is automatic: you cannot look at a familiar word and not read it. Naming a colour is something you do less often and more deliberately. When both arrive at once and point to different answers, the automatic one has to be held back before the right answer can be given, and holding it back takes time. Occasionally it fails, and you say the word instead of the colour.

That is what makes the test interesting for attention. The extra time is a measure of how much effort it takes to stay with the task you were given while something more familiar pulls the other way. Psychologists call this kind of control inhibition, or selective attention.

It also means the effect depends on reading the words fluently, without effort. That is why our test shows the colour words in the language of the page, where they are read most automatically.

Modern versions show one word at a time on a computer and time each answer, made with a key for each colour. In a study published in 2018, undergraduate students answered hundreds of words in a session. Their answers to conflicting words were on average 78 ms slower than to matching words, and individuals were spread around that by about 38 ms^[2].

Almost everyone shows the effect. That makes the Stroop test one of the most reliable demonstrations in psychology: run it on any group and conflicting words will be slower.

The same study asked a harder question: does a person's Stroop score stay the same from one day to another? The students were tested twice, 3 weeks apart. The agreement between the two sessions was 0.60 in one study and 0.66 in a second, on a scale where 0.82 was the best any of their seven tasks reached^[2].

The authors called this the reliability paradox. A task becomes famous because it works for nearly everyone, which means people differ little in it; and when people differ little, measurement noise is large compared with the real differences between them. So the Stroop effect is very dependable for a group, and much less dependable as a way to rank one person against another^[2].

• A Stroop score is a good way to see the effect in yourself, and to watch it change with practice or tiredness.
• It is a poor way to compare your attention with someone else's on the basis of one short run.
• It says nothing about how clever you are or about your health; it is for curiosity and self-reflection.

Every answer in a Stroop test, even to a matching word, is slower than a plain reaction. In a simple reaction test there is one signal and one button; here you have to see the colour, pick the right key out of several and press it. Choosing takes time, which is why the times in a Stroop test are measured in larger steps than a reaction test's, and why the score is the difference between two kinds of word rather than a single time.

Choices are also where age shows most. A survey of 7,130 adults in the United Kingdom found that simple reactions changed little until about 50, while reactions with a choice slowed across the whole adult range^[3]; a long study in Baltimore found a reaction with a decision slowing about three times as fast with age as a simple one^[4]. An older adult can therefore expect slower Stroop times overall, which is one more reason the test compares the two kinds of word instead of looking at speed alone.

Our Stroop test begins with 8 practice words that do not count, then times 48 words, half of them in a conflicting colour. Your score is the difference between your middle time for conflicting words and your middle time for matching ones, together with how many you answered correctly. The middle time is used because a single slow answer moves it less than it moves an average.

Because the test is short, your score will move from run to run. Take it a few times and look at the usual size of your effect rather than a single number.

Why every Stroop answer is a choice reaction, and how reaction times are measured in general, is explained in how reaction time is measured.