Stroop Effect Test — Standard Scores & Reference Values
What do Stroop test numbers actually look like in the published literature? This page collects the reference values — the original experiment, typical interference costs in milliseconds, and the effects of age and practice — with the source attached to every figure.
What the Stroop effect is — and how big it is
The Stroop effect is the slowdown that happens when a color word fights its own ink color. In the original Stroop experiment (J. Ridley Stroop, 1935), participants naming the ink colors of 100 conflicting color words needed 110.3 seconds on average, versus 63.3 seconds to name the same colors printed as solid squares — a 74.3% interference cost. Reading the same words aloud was barely affected (+5.6%, not statistically reliable), which is why the task measures the word’s interference on color naming, not the reverse. Source: https://psychclassics.yorku.ca/Stroop/
On modern button-press versions (like the test on this site), the effect appears in single-trial reaction times rather than stopwatch totals, and published values for healthy adults land in a consistent range: 589.6 ms on congruent trials versus 661.2 ms on incongruent trials — an interference cost of about 72 ms — in an fNIRS study (https://worldscientific.com/doi/pdf/10.1142/S0219519421400224?download=true); and a mean difference of 106.3 ms across 81 participants in a recent preprint (https://www.researchsquare.com/article/rs-10287577/latest.pdf). Expect your own interference cost to sit roughly in the tens to low hundreds of milliseconds.
Two forces move your score more than anything else. Age: interference follows a U-shape across the lifespan — highest in childhood, lowest in young adults, rising again in older age (Comalli, Wapner & Werner 1962, as summarized in Ferrand et al. 2020: https://hal.science/hal-02324227/file/Ferrand-et-al-2020-DevScience-HAL.pdf); a normative study of 1,856 healthy Dutch speakers aged 24–81 likewise found performance declines with age and varies with education (https://www.science.gov/topicpages/c/color+word+test.html). Practice: in Stroop’s own 1935 practice study, eight days of 200 daily trials cut time on the interference task by 33.9% — a large gain, yet the effect shrank without disappearing (https://psychclassics.yorku.ca/Stroop/). For the theory behind all of this, MacLeod’s 1991 integrative review remains the standard reference (https://doi.org/10.1037/0033-2909.109.2.163).
Reference values, each with its source
| Measure | Reference value | Source |
|---|---|---|
| Original card test (1935), per 100 items | 63.3 s naming plain colors → 110.3 s naming conflicting ink (+74.3%) | https://psychclassics.yorku.ca/Stroop/ |
| Computerized interference cost (incongruent − congruent RT) | ≈72 ms (fNIRS study); 106.3 ms mean (n = 81, preprint) | https://worldscientific.com/doi/pdf/10.1142/S0219519421400224?download=true · https://www.researchsquare.com/article/rs-10287577/latest.pdf |
| Practice effect (8 days × 200 trials, 1935) | −33.9% time on the interference task; the effect shrank but did not disappear | https://psychclassics.yorku.ca/Stroop/ |
| Age trend | U-shape: highest in childhood, lowest in young adults, rising again in older age; adult norms (n = 1,856, ages 24–81) confirm decline with age and an education effect | https://hal.science/hal-02324227/file/Ferrand-et-al-2020-DevScience-HAL.pdf · https://www.science.gov/topicpages/c/color+word+test.html |
| Background theory | The standard integrative review of the first 50 years of Stroop research (1935–1990) | https://doi.org/10.1037/0033-2909.109.2.163 |
| Correct answers in 45 s on this site | No published norm exists for 45-second button-press rounds | SchulteLab user aggregate (beta — sample still accumulating) |
Honesty note: computerized button-press versions (including ours) systematically differ from the classic 100-item card version, so compare like with like. There is no official 45-second scoring band — where no published norm exists, we label our own user aggregate as beta rather than invent numbers.
Age anchors from the card version — paper-based, spoken, NOT comparable with this web test
The only large age-stratified Stroop samples in the literature come from the classic card version (read the ink color aloud, 100+ items, stopwatch scoring). They show the expected age trend — interference-card time rises steadily after mid-life — but they are PAPER-AND-PENCIL SPOKEN anchors: do not compare them against any web touch-screen or button-press score, including the assessment on this site.
| Study (card version, spoken) | Age group | Interference-card completion time | Source |
|---|---|---|---|
| Hankee et al. 2016 (N = 3,362, Framingham) | <35 years | 20.3 s | https://pmc.ncbi.nlm.nih.gov/articles/PMC4946576/ |
| Hankee et al. 2016 (N = 3,362, Framingham) | 35–44 years | 22.2 s | https://pmc.ncbi.nlm.nih.gov/articles/PMC4946576/ |
| Hankee et al. 2016 (N = 3,362, Framingham) | 45–54 years | 25.3 s | https://pmc.ncbi.nlm.nih.gov/articles/PMC4946576/ |
| Hankee et al. 2016 (N = 3,362, Framingham) | ≥55 years | 28.6 s | https://pmc.ncbi.nlm.nih.gov/articles/PMC4946576/ |
| Tremblay et al. 2016 (N = 646) | 47–87 years | 35.32 (12.93) s — mean (SD) | https://pmc.ncbi.nlm.nih.gov/articles/PMC5859918/ |
Card-version anchors only — not norms for this site. In Hankee 2016 the interference ratio (T3/T1) also rises from 1.7 to 2.2 across those age groups, the clearest published age gradient for the Stroop effect itself.
Why no published touch-screen norm exists
Three findings explain why nobody can honestly publish age norms for a web touch-screen Stroop test yet. First, the versions do not transfer: Penner et al. 2012 found that scores on the card version and a computerized version did not correlate (DOI: 10.1080/13854046.2012.713513), so card norms cannot be re-labeled as computerized norms. Second, the effect itself shrinks on touch screens: Fennell et al. 2019 measured only ~15–17 ms of interference on a 4-choice touch version versus ~114 ms on the spoken version (https://pmc.ncbi.nlm.nih.gov/articles/PMC6923799/). Third, the device moves the baseline: Passell et al. 2021 found mobile devices run 50–200 ms slower than desktop on the same tasks (https://pmc.ncbi.nlm.nih.gov/articles/PMC8568735/) — a phone and a laptop are different instruments.
Our position: rather than borrowing norms the literature says do not transfer, this site records the device class (touch vs mouse) with every anonymous score and is accumulating a device-stratified community distribution — the live strip under the reference table above is that growing sample. When it matures, we will publish our own openly-described reference values instead of someone else’s.
Take or print the test
Measure your own interference cost in 45 seconds, or print fresh color-word cards for offline use. Open the printable Stroop cards →
Frequently asked questions
What is a good score on a Stroop effect test?
There is no official cutoff. In published button-press studies with healthy adults, incongruent trials run about 70–110 ms slower than congruent ones (≈72 ms in the fNIRS study and 106.3 ms in the 81-person preprint linked above), so an interference cost in that range is typical. Judge yourself by the trend in your own history, not a single round.
Does age change Stroop test results?
Yes. Interference follows a U-shaped curve: highest in childhood, lowest in young adults, and rising again in older age (Comalli et al. 1962, summarized in the Ferrand et al. 2020 paper linked above). A 1,856-person normative study of adults aged 24–81 also found performance declines with age and is influenced by education level.
Can practice improve my Stroop score?
Partly. In Stroop’s 1935 practice experiment, eight days of 200 daily trials cut time on the interference task by 33.9% — but the effect itself shrank without vanishing. Expect to get measurably faster at the task (near transfer); treat any claim of broader cognitive benefits as unproven.
Why is there no norm for the 45-second correct count?
Because none exists in the published literature: standardized Stroop versions use 100-item cards or millisecond reaction times, not 45-second button-press rounds. Rather than invent a number, we label our own user aggregate as beta while the sample accumulates. The sourced values in the table are the honest reference points.
Can I trust the age percentiles on sites like strooptester?
Treat them as order-of-magnitude orientation only. Those sites apply card-version (spoken, paper) norms to button-press or touch-screen tests, and the published literature does not support that transfer: card and computerized scores do not correlate (Penner et al. 2012), touch versions show a much smaller interference effect (Fennell et al. 2019), and mobile devices run 50–200 ms slower than desktop (Passell et al. 2021). An age percentile computed across versions mixes different instruments.