Methodology: how the free dyslexia test is scored
Scoring last changed
The free Dyslexia Steps test is a screening: 6 short games for children (about 8 minutes) or 15 questions and 2 tasks for adults, scored in the browser. Each game gives a score from 0 to 100% against a rough expectation for the age group. Below 60% a skill is marked "Needs practice", and when 2 or more of the 4 core reading skills fall below that line, the result suggests asking for a professional assessment. These expectations and cut-offs are cautious choices, not norms from a tested sample, and the test has not been validated against a professional assessment. Every change to the scoring is dated on the changelog.
What the test measures, and what it does not
The child test looks at 6 skills that reading is built on: hearing sounds in words, matching letters to sounds, reading made-up words, naming speed, telling look-alike letters apart and remembering spoken sequences. The result is a profile of those skills from one sitting, with a strongest and a focus skill.
It is a screening, not a diagnosis. It does not tell you whether a child has dyslexia. It does not measure reading comprehension, spelling or writing; intelligence, attention (for example ADHD), vision or hearing; or maths skills (our sister site dyscalculia.io has a separate number-skills screening).
The 6 games in the child test
Every child sees the same items in the same order; the test does not adapt. Each listening game speaks its instruction and the item first, and answers open only after the audio has played; answering slowly does not lower those scores. Colour Race measures speed on purpose, and Mirror Letters has a time limit per round.1,2,3,4,5
| Game | Skill | Items | Scored by |
|---|---|---|---|
| Odd Sound Out1 | Hearing sounds in words | 10 spoken word triples: which one does not rhyme, then which starts with a different sound | share of correct answers |
| First Letter2 | Letters and sounds | 10 spoken words, 4 letters to choose from | share of correct answers |
| Alien Words3 | Reading new words | 10 spoken made-up words, 4 spellings to choose from | share of correct answers |
| Colour Race4 | Naming speed | 40 coloured dots (5 rows of 8, 5 colours) named aloud; a parent presses Start and Stop and marks errors | seconds per colour against the age expectation, minus 5% per error |
| Mirror Letters | Spotting letter details | 8 rounds: tap every target letter in a grid of 16 (4–5 targets, 12 seconds) | share of targets found minus share of wrong taps |
| Number Echo5 | Memory for sounds | 11 spoken digit sequences growing from 2 to 7 digits; stops after two misses at the same length | longest sequence repeated, against the age expectation |
How a game becomes a score
For the four accuracy games, the result is divided by the accuracy we expect for the age group and capped at 100%. A 7-year-old with 6 of 10 correct, where we expect 75%, scores 80%.
Colour Race compares seconds per colour with a typical time for the age group: at the typical speed or faster the score is 100%, at twice the typical time it is 0, and each error the parent marks takes off 5% ("1–2 errors" counts as 1.5, "3 or more" as 4).
Number Echo scores the longest sequence repeated correctly: the typical length for the age group scores about 91%, and a sequence 3 digits shorter than typical scores 0.
| Age group | Expected accuracy | Colour Race: typical seconds per colour | Number Echo: typical length |
|---|---|---|---|
| Child, 5–6 | 60% | 1.4 | 4 |
| Child, 7–8 | 75% | 1.05 | 5 |
| Child, 9–10 | 85% | 0.85 | 5 |
| Child, 11–12 | 90% | 0.75 | 6 |
| Teenager | 92% | 0.65 | 6 |
| Adult | 95% | 0.55 | 7 |
What the result says
Each skill is shown as a bar with one of three labels: "Strong" from 85%, "On track" from 60% up to 85%, and "Needs practice" below 60%. The strongest skill is the highest of the 6 scores and the focus skill the lowest; if scores tie, the one listed first wins.
The note under "What this means" changes when 2 or more of the 4 core reading skills (Hearing sounds in words, Letters and sounds, Reading new words, Naming speed) are below 60%. It then suggests sharing the result with the child's teacher and asking about a professional assessment. The two other skills are shown but never trigger that note, because the evidence linking them to reading is weaker.
If you create an account, the program's daily plan starts from the focus skill. The result itself is the same with or without an account.5
The adult check
The adult path has no games for children. It asks 15 questions about reading and spelling now and at school, each answered on a 4-point scale from "Never" to "Almost always"; the score is 1 minus the average answer divided by 3, so more difficulty gives a lower score. The questions are our own, modelled on the approach of published adult reading-history questionnaires, not copied from one and not validated.
Sounds Like a Word shows 12 pairs of made-up spellings and asks which sounds like a real word. Its score is 70% accuracy and 30% speed: a median answer time of 1 second or less counts as full speed and 3.5 seconds or more as none, because adults with dyslexia are often accurate but slow. Colour Race works as for children, timed by the adult, with the adult expectation.
The note says the answers show several signs common in adults with dyslexia when the questionnaire score is below 60%, or when both tasks are below 60%.6,3,4
Where the expectations and cut-offs come from
We have no norms. The expected accuracies, typical naming times and typical sequence lengths in the table above are rough expectations we set for each age group, guided by research showing that these skills grow with age; they were not taken from a normed test and they were not fitted to a sample of children with and without dyslexia.
The 60% line and the rule of 2 core skills are deliberately cautious: the aim is to point out when a follow-up is worth asking for, not to sort children into groups. Because the test has not been compared with professional assessments, we publish no accuracy, sensitivity or specificity figure for it. If the expectations or cut-offs change, the old and new values will be on the changelog.4
Languages and expert review
The test runs in 6 languages. Each language has its own test content (words, made-up words and audio) written for how that language is spelled, not translated, because the same skill looks different in English and in more regularly spelled languages. Every language has the same number of items and the same scoring.
The test content in every language, English included, is a draft that has not yet been checked by a reading specialist. When a specialist reviews it, their name and title will appear on the about page, and the review will be dated on the changelog. The audio is synthetic speech.7,8
Limits of an online screening
Treat the result as one observation, not a measurement you can rely on alone.
One sitting at home
Tiredness, distraction, a quiet speaker or a hard day all move the scores.
A parent keeps the time
In Colour Race, a parent presses Start and Stop, so the timing depends on the parent.
It can miss, and it can over-flag
The test can miss a child who needs support and can flag a child who does not. A teacher's concern matters more than any online result.
Teenagers and adults
Teenagers use the child games with the 13+ expectations; the adult path is a questionnaire with two tasks, not a full assessment.
What the reading games record
The paid program has 156 levels across 7 worlds of games. A level earns 3 stars from 85% correct, 2 stars from 70% and 1 star from 50%, and 1 star opens the next level. Stars describe accuracy on that level only.
Words and letters come back for review on a spaced schedule: after each correct answer the gap grows (1, 2, 4, 8, 16 days), after a wrong one the item starts again and is due at the next session, with up to 8 review items a day. The parent area shows an answer-speed trend that compares the child's most recent timed sessions (up to 5) with the same number before them, game by game; changes within 10% count as "about the same".
None of this is compared with other children: there are no norms, percentiles or age scores in the program, and progress in the games is not a test result.
The program is built on the approaches with the strongest research support for children who find reading hard: phonological awareness, systematic phonics and decoding, plus spaced review. The visual and memory worlds are warm-ups, and we claim no reading benefit for them. No study has tested the Dyslexia Steps program itself; the pages describe what the games practise and what research says about those skills, not a measured result of our program.2,1,3,9
Sources the site cites
Factual sentences on the information pages link to the site's reference list of 146 sources: 52 with a PubMed record, 12 with a DOI only, and 82 linked by URL (official statistics, guidelines, organisation pages and articles outside PubMed). The figures on the dyslexia statistics page each carry the year of their data and are also published as a downloadable file.
Sources
- Melby-Lervåg M, Lyster SA, Hulme C (2012). Phonological skills and their role in learning to read: a meta-analytic review. Psychol Bull, 138(2), 322-52.
- Galuschka K, Ise E, Krick K, Schulte-Körne G (2014). Effectiveness of treatment approaches for children and adolescents with reading disabilities: a meta-analysis of randomized controlled trials. PLoS One, 9(2), e89900.
- Norton ES, Wolf M (2012). Rapid automatized naming (RAN) and reading fluency: implications for understanding and treatment of reading disabilities. Annu Rev Psychol, 63, 427-52.
- Peterson RL, Pennington BF (2012). Developmental dyslexia. Lancet, 379(9830), 1997-2007.
- Lefly DL, Pennington BF (2000). Reliability and validity of the Adult Reading History Questionnaire. J Learn Disabil, 33(3), 286-96.
- Seymour PHK, Aro M, Erskine JM (2003). Foundation literacy acquisition in European orthographies. Br J Psychol, 94(2), 143-74.
- Ziegler JC, Goswami U (2005). Reading acquisition, developmental dyslexia, and skilled reading across languages: a psycholinguistic grain size theory. Psychol Bull, 131(1), 3-29.
- Cepeda NJ, Pashler H, Vul E, Wixted JT, Rohrer D (2006). Distributed practice in verbal recall tasks: a review and quantitative synthesis. Psychol Bull, 132(3), 354-80.