Mental health

What the RAADS-R, AQ, and CAT-Q Can and Can't Tell You

Save

Three questionnaires dominate adult self-screening: the RAADS-R, the AQ, and questionnaires about camouflaging like the CAT-Q. Each was designed for a different job, and none was designed to be scored by the person taking it on a website. This explains what each one measures, where its numbers came from, what a score above the cutoff does and does not mean, and what to do with the result.

Last updated: July 2026

Talk to a clinician

Gale can help you find a clinician in your state and request a visit.

Find care →

Are online autism tests like the RAADS-R accurate?

The instruments themselves are accurate in the setting that produced their numbers. The RAADS-R's international validation reported 97% sensitivity and 100% specificity across 779 people, 201 of them autistic, all assessed against full diagnostic criteria 1. Those figures describe a scale administered inside a clinical workup. A free version of the same questions, scored by you at midnight, is not the thing that was studied.

Sensitivity is how often a test flags the people who do have the condition. Specificity is how often it correctly passes over the people who do not.

Both numbers depend on who is being tested. Online, the people typing the name of a screen into a search bar overwhelmingly already suspect the answer. In a group leaning that far in one direction, a high score carries less information than the headline sensitivity figure suggests.

The accuracy figures belong to the clinic that generated them, not to the copy of the questionnaire on a website.

None of that makes adult self-screening tools worthless. It makes them a starting point rather than a verdict.

What the RAADS-R was designed to do

The RAADS-R is an 80-question scale covering social relatedness, circumscribed interests, language, and sensory-motor experience, and it was built as an aid to diagnosis rather than a diagnosis in itself 1. It was validated on adults who had already been assessed with structured diagnostic interviews and standardized IQ testing, which is the context its numbers belong to.

Two other findings from that study are worth knowing. Its test-retest reliability was r = .987, meaning people who took it twice tended to land in close to the same place both times — the scale is stable 1. And it agreed with an established adult measure of social responsiveness 95.59% of the time 1. A stable, agreeing instrument is a good instrument. Neither property turns it into a diagnosis.

Its authors describe something a clinician uses while assessing an adult. That is the job it does well. Where a website prints a pass/fail threshold beside your total, it helps to remember that thresholds are set to sort a research sample into groups, not to rule on the person reading the screen.

What the AQ measures, and what its cutoff was for

The Autism-Spectrum Quotient is a 50-item self-report questionnaire across five areas — social skill, attention switching, attention to detail, communication, and imagination — scored from 0 to 50, with higher totals meaning more autistic traits 2. It was designed to quantify traits in adults of normal intelligence, and its original validation compared autistic adults against randomly selected controls, Cambridge University students, and mathematics olympiad winners 2.

80% of the autistic adults in that study scored 32 or above, against 2% of randomly selected controls 2.

The group averages were 35.8 (SD 6.5) for the autistic participants and 16.4 (SD 6.3) for controls 2. Those distributions sit far apart, but they overlap at the edges: some controls scored high, and one in five autistic participants scored below the line. That overlap is ordinary screening behaviour, and it is why a total of 30 or 33 says much less about one person than the gap between the averages suggests.

Two details matter for adults who suspect they were missed as children. Among the controls, men scored slightly higher on average than women 2 — a pattern that shaped decades of assumptions about who gets referred. And the instrument was developed for adults of normal intelligence 2, a narrower group than the spectrum as a whole.

Where camouflaging questionnaires like the CAT-Q fit

The CAT-Q — the Camouflaging Autistic Traits Questionnaire — asks about a different thing from the RAADS-R and the AQ. It measures effort: the compensating, masking and assimilating people do to pass as non-autistic in social settings. It is not a substitute for a trait screen, and a high camouflaging score is not evidence of autism on its own.

That distinction matters most for the group most likely to be reading this page: adults who were never referred as children because they were quiet, compliant, or academically fine. A trait questionnaire asks what you do. A camouflaging questionnaire asks what it costs you to do it. Two people can produce the same AQ total and be living entirely different lives underneath it.

None of this is a shortcut around an assessment. But understanding the autism spectrum as a range of support needs rather than a severity ladder tends to explain a confusing set of scores better than any single number does.

Why a high score is not a diagnosis

Screening and diagnosis are two different steps. A screening tool sorts people into worth-a-closer-look and probably-not, and a positive screen is an indication for a comprehensive evaluation rather than an answer 3. Autism is identified from behaviour, developmental history, and direct observation — a questionnaire supplies one input to that, not the conclusion.

A positive screen is a reason to be evaluated. It is not a result.

A proper adult autism assessment is built to do what a form cannot. Guidance for assessing adults describes a comprehensive, multidisciplinary process that includes taking a developmental history — what you were like before you learned to compensate, gathered where possible from someone who knew you then 4. Autism is a developmental condition whose features are typically recognisable in the first two years of life and whose support needs can continue across the lifespan 5. An assessment therefore spends as much time on your childhood as on your present, which is exactly the material a self-administered questionnaire has no access to.

What changes when the questionnaire lives on a website

Four things change: nobody checks whether you understood the items the way the authors meant them, nobody rules out the other explanations that produce similar answers, the sample answering online is people who already suspect they are autistic, and the result goes nowhere — it does not enter a medical record and no clinician has to act on it.

  • Nobody checks the reading. In a clinic, an item you interpret differently from how it was written gets caught in conversation. Alone with a form, it does not.
  • Nobody rules out the alternatives. ADHD, social anxiety, trauma, and long-running depression can all push answers in the same direction.
  • The sample is self-selected. Almost everyone taking an online autism screen arrived already suspecting the answer, which is not the population the instrument's published numbers came from.
  • The result goes nowhere. It does not enter a record, no clinician has to act on it, and it establishes nothing with an employer or a school.

A telehealth autism assessment is a different animal entirely. A clinician taking a history over video is doing the assessment; an automated quiz is not, whatever it is priced at. The useful question is who reads your answers and what qualification they hold.

What a screening score is genuinely good for

It is good for making an appointment possible. A printed score, plus a page of your own examples, turns "I think something is different about me" into a concrete request a clinician can act on. Adults who reach a diagnosis late often describe it as reframing decades of experience rather than delivering news 6.

What helps most is what you bring alongside the number. Concrete material — what a day of open-plan work costs you, what happens in the two hours after a party, what teachers wrote about you at seven — gives an assessor something a total score cannot supply.

Nothing you find in a questionnaire obliges you to do anything with it. A result can sit unused for a year and lose none of its meaning.

It is also reasonable to decide the score is enough for now. The self-identification vs diagnosis question is genuinely open in the adult autistic community, and the adult diagnosis tradeoffs — cost, waiting lists, what a formal record does and does not change at work — are worth weighing before booking anything.

If the score doesn't match what you expected

A score below a threshold does not close the question, and a score above it does not settle it. Overlapping experiences — social anxiety, ADHD, trauma, long-running depression — can produce similar answers on a self-report form. Sorting them out is what a comprehensive assessment does, which is why it takes a developmental history and, where possible, an account from someone who knew you as a child 4.

Telling anxiety vs autism apart is the most common version of this. Both can produce avoidance of social situations, and both can make a screening item feel unmistakably true. The distinction an assessor works with is whether the difficulty has been lifelong and present across settings, or whether it arrived at a particular point and attaches to particular fears. That is a history question, not a questionnaire question.

There is also the paid-report problem. Some sites charge for a score, a written report, or a certificate. It is worth knowing what the money buys before paying: an automated number, a document generated from your own answers, or an evaluation carried out by a licensed clinician who takes a history and puts their name on it. Only the last is an assessment.

Common questions

The questions are usually the same, but the conditions are not. In the validation study the scale was administered as part of a clinical assessment, alongside structured diagnostic interviews. Online, nobody clarifies an item you read differently from how it was written, and nobody weighs your answers against a developmental history. Same questionnaire, different instrument.

No clinician diagnoses from a questionnaire score alone. A score can open the conversation and shorten it: it gives the person assessing you a starting point and shows you arrived with something specific rather than a vague worry. The diagnosis itself comes from a full assessment, which looks at childhood as closely as it looks at the present.

Not on its own. Self-report screens ask you to judge your own behaviour, which is exactly the judgement that years of masking can distort. Adults who learned early to perform an expected version of themselves often answer as that version. A score below a threshold is one piece of information, and it does not close the question.

The question has no clean answer, because the published accuracy of each instrument was measured in a clinical setting rather than online, and the studies used different samples and different comparison groups. Taking two screens and coming away with two different impressions is common, and not a contradiction: they measure overlapping but genuinely different things.

Many people find it easier to open with the questionnaire than with the feeling underneath it, and a printed result is a concrete thing to hand over. It also helps to bring examples: what a day at work costs you, what happens after a social event, what you were like at seven. The examples usually carry more weight than the number.

Related

Say it back

How would you explain this to someone you love?

Two or three sentences, just as you’d say it. Gale reflects back what you focused on — a mirror, not a quiz.

If things feel heavy, a person is available anytime — call or text 988.

Talk to a clinician

Gale can help you find a clinician in your state and request a visit.

Find care →

When a screening result needs more than a follow-up appointment

  • —Thoughts of ending your life, or a plan taking shape, in the days after reading about autism or taking a screen
  • —A collapse in functioning lasting more than a few days — unable to work, wash, eat, or speak — after a long stretch of sustained masking
  • —New panic attacks, or a shutdown severe enough that you cannot leave the house or answer the door
  • —Drinking or using substances noticeably more to get through social or work demands

If you are thinking about ending your life, call or text 988 (the Suicide and Crisis Lifeline) or text 741741. If you have already acted on those thoughts, call 911 or go to the nearest emergency department.

This page explains what adult autism screening questionnaires measure and where their numbers come from. It is general information, not a diagnosis and not medical advice. No questionnaire — taken here or anywhere else — can establish whether you are autistic; a licensed clinician who takes a full developmental history is the only route to that answer.

References

  1. 1.Ritvo RA, Ritvo ER, Guthrie D, et al. (2011). The Ritvo Autism Asperger Diagnostic Scale-Revised (RAADS-R): a scale to assist the diagnosis of Autism Spectrum Disorder in adults: an international validation study. Journal of Autism and Developmental Disorders, 41(8), 1076-89. doi:10.1007/s10803-010-1133-5 ✓The RAADS-R is an 80-question scale spanning social relatedness, circumscribed interests, language and sensory-motor domains, validated in 779 adults (201 with ASD, 578 comparisons) who also met DSM-IV-TR criteria with ADI/ADOS and standardized IQ testing; reported sensitivity 97%, specificity 100%, test-retest reliability r = .987, and 95.59% concurrent validity with the adult Social Responsiveness Scale; positioned as a diagnostic adjunct rather than a stand-alone diagnosis.
  2. 2.Baron-Cohen S, Wheelwright S, Skinner R, et al. (2001). The autism-spectrum quotient (AQ): evidence from Asperger syndrome/high-functioning autism, males and females, scientists and mathematicians. J Autism Dev Disord. 2001 Feb;31(1):5-17. doi:10.1023/A:1005653411471 ✓The AQ's structure and original validation: 50 self-report items across five subscales (social skill, attention switching, attention to detail, communication, imagination), scored 0-50 with higher totals indicating more autistic traits; 80% of the AS/HFA group scored 32 or above versus 2% of randomly selected controls; group means of 35.8 (SD 6.5) versus 16.4 (SD 6.3); males scoring slightly higher than females among controls; and its development for adults of normal intelligence, with comparison groups including Cambridge students and Mathematics Olympiad winners.
  3. 3.Centers for Disease Control and Prevention (2024). Screening and Diagnosis of Autism Spectrum Disorder. CDC — Autism Spectrum Disorder (ASD). linkThat identification is a two-step process — developmental screening followed by a comprehensive diagnostic evaluation — and that a positive screen indicates the need for further evaluation rather than constituting a diagnosis.
  4. 4.National Institute for Health and Care Excellence (NICE) (2021). Autism spectrum disorder in adults: diagnosis and management (CG142). NICE Clinical Guideline (via NCBI Bookshelf). link ✓That a comprehensive adult autism assessment is multidisciplinary and includes taking a developmental history, including information from an informant who knew the person in childhood where that is available — the material a self-administered questionnaire cannot supply.
  5. 5.National Institute of Mental Health (2024). Autism Spectrum Disorder. National Institute of Mental Health (NIMH). link ✓That autism is a developmental condition whose features are typically recognisable in the first two years of life and whose support needs can continue across the lifespan into adulthood.
  6. 6.Huang Y, Arnold SR, Foley KR, Trollor JN (2022). Late diagnosis of autism: exploring experiences of males diagnosed with autism in adulthood. Current Psychology. link ✓That adults diagnosed in adulthood commonly describe the diagnosis as reframing long-standing experiences and reshaping identity, rather than as new information about themselves.

6 sources, numbered by first appearance. General health information, not medical advice. AI-assisted editorial content — citations link their sources. Editorial policy

← Back to the health library