Autism · 80 items (not reproduced) · No published maximum · Ritvo, Ritvo, Guthrie and colleagues (2011), an international validation study across nine centres.
Licensed instrument
RAADS-R: scoring, cutoffs & interpretation
An 80-item clinician-administered measure of autistic traits in adults, with a published threshold of 65.
The RAADS-R is an 80-item measure of autistic traits in adults of average or above-average intelligence, built around the DSM-IV-TR criteria and organised into four clinical subscales: social relatedness (39 items), circumscribed interests (14 items), language (7 items) and sensory motor (20 items). Sixty-four of the items are symptom-based. The remaining sixteen are non-symptom "normative" items describing ordinary experience, and these are reverse-scored; they are included to interrupt a straight-line response pattern in which a respondent simply works down one column.
Its distinguishing feature is that every item asks the respondent to place an experience in developmental time rather than to agree or disagree with a statement. The four response options separate something true now and in childhood, true only now, true only before the age of 16, and never true, scored 3 to 0 respectively and reversed for the sixteen normative items. That framing matters because adult autism assessment turns on whether traits were present from early life rather than on how someone presents today.
02 - Origin & purpose
Where it comes from.
The RAADS-R was developed by Riva Ariella Ritvo, Edward Ritvo and colleagues as a revision of the earlier RAADS, and was validated across nine centres in North America, Europe and Australia before publication in the Journal of Autism and Developmental Disorders in 2011. The validation sample was 779 people: 201 with an autism spectrum diagnosis (66 autistic disorder, 135 Asperger's disorder) and 578 comparison participants, of whom 302 were adult psychiatric outpatients with a non-autism Axis I diagnosis. Autistic participants had to have a full-scale IQ of 80 or above and an ADOS module 4 score in the autism range.
The developers designed the instrument for clinician administration and warned explicitly against the use it is now best known for: "It is designed to be administered by clinicians in a clinical setting. It is not intended to be a mail in or an online screening instrument." In the validation study a clinician stayed with each participant while they answered, to clarify questions and to check that responses were entered in the correct column. The RAADS-R has since become one of the most widely self-administered autism questionnaires online, and much of the confusion about what its scores mean follows from that gap between how it was designed and how it is used.
03 - Scoring & cutoffs
How scoring works.
The 80 items are each scored 0 to 3, with the sixteen asterisked normative items reversed. A total of 65 or above is the published threshold, described by the authors as "consistent with a clinical diagnosis of ASD", against sensitivity of 97% and specificity of 100% in the validation sample.
Two points of precision are routinely got wrong elsewhere. First, the threshold is 65, not 64: the authors' own later article in Autism Spectrum News says "64 or higher", which contradicts the threshold given in their journal paper, and the peer-reviewed paper is the figure to use. Second, the paper publishes no theoretical maximum score. It reports observed ranges only - 44 to 227 among autistic participants and 0 to 65 among comparison participants - so no maximum total is given here: a figure derived by multiplying items by the highest response value is arithmetic rather than a published result.
Six autistic participants (3%) scored below 65 in the validation sample, and the authors are explicit that clinical judgement overrides a sub-threshold score.
04 - Validation evidence
How well it performs.
Test-retest reliability was reported as r = 0.987 in 30 participants over a mean interval of one year. Internal consistency was reported for each of the four clinical subscales: alpha 0.923 for social relatedness, 0.905 for sensory motor, 0.903 for circumscribed interests and 0.789 for language. No alpha for the total scale is reported anywhere in the paper, so a total-scale alpha quoted elsewhere did not come from the source. Concordance with informant report on the SRS-A across 69 first-degree relatives was 95.59% - though the paper itself gives that figure as 95.59% in its results and 95.95% in its limitations section.
An exploratory factor analysis produced four factors that do not match the clinical subscales: social relatedness, circumscribed interests, sensory motor and a social anxiety factor, with the language construct not emerging at all. Inter-rater reliability is not reported. Because the ADOS and ADI were used as inclusion criteria rather than as comparators, no correlation between the RAADS-R and the ADOS is published.
Informant-completed and developmental. Useful when a parent or carer can report on early childhood.
ADOS-2
structured observation
40-60 min
Not a self-report measure. The standardised observational assessment the questionnaires point towards.
06 - When to use it
Right tool, wrong tool.
Read the RAADS-R as one strand of an adult autism assessment alongside a developmental history and clinical interview, ideally sitting with the person while they complete it so that items can be clarified. Where a brief screen is what is actually wanted, the RAADS-14 Screen or the AQ-10 are the practical choices; both are short and freely reproducible.
Reach for it when
-A detailed developmental self-report is wanted as part of a full adult autism assessment
-A clinician can sit with the person while they complete it
-Traits need to be placed in developmental time rather than described only as they present now
Reach for something else when
-As a brief screen - the RAADS-14 Screen or AQ-10 are shorter and freely usable
-As a stand-alone diagnostic instrument
-Unsupervised online self-administration, which the developers advise against
-In adults with intellectual disability, where it has not been validated
07 - Confidence & precision
Reading the score with care.
The specificity of 100% reported in the original study has not survived replication in the populations where the question is actually asked. Jones and colleagues (2021) gave the RAADS-R to 50 consecutive referrals to an NHS adult autism service: 49 of the 50 scored above 65, and 17 (34%) went on to receive a diagnosis. Sensitivity was 100%, specificity 3.03%, positive predictive value 34.7%, and the area under the curve was 0.45 - statistically no better than chance. They concluded that it "lacks predictive validity and is not a suitable screening tool" when self-completed before assessment. Their own caveat should be given equal weight: they administered it by post, with the person answering alone, contrary to the developers' instructions.
Other replications point the same way. Sizoo and colleagues (2015) found that non-autistic psychiatric controls scored on average well above the threshold and proposed a higher cut-off of 98. Brugha and colleagues (2020), in 738 adult mental health service users, found that more than half exceeded 65 and advised against using the threshold at that level, suggesting a figure closer to 120 to 126. Picot and colleagues (2021), validating the French version in 305 people, found a false-positive rate above 50% in the psychiatric group. For balance, Sturm and colleagues (2024) found the instrument psychometrically sound and unidimensional, without bias by age, gender or diagnosis, while still concluding that it is a screener rather than a diagnostic instrument.
The practical consequence is straightforward. A RAADS-R score above 65 in someone who already has a psychiatric diagnosis carries very little information, and the people most likely to have completed it online are exactly that group. A high score is a reasonable prompt to seek assessment. It is not, on its own, evidence of autism.
08 - Limitations
What it cannot tell you.
The RAADS-R was validated only in adults with a full-scale IQ of 80 or above, and under DSM-IV-TR, before the DSM-5 criteria for autism were introduced. There were significant differences between the nine study centres, whose samples ranged from 7 to 53 participants. As a self-report measure it is vulnerable to masking: the authors note that each of the six autistic participants who scored below the threshold was described by family members as invested in appearing "as normal as they could be". It is an adjunct to assessment rather than a diagnostic instrument, and there is a substantial mismatch between its clinician-administered design and the online self-administration through which most people now encounter it. The paper also contains no competing-interest or funding declaration, despite two of its authors being the instrument's developers.
09 - Licensing, explained
How licensing works.
The RAADS-R was published in an open-access article distributed under a Creative Commons Attribution-NonCommercial licence, which permits non-commercial use, distribution and reproduction with credit. The instrument itself appears as an appendix to that article and is covered by the same licence, so the items cannot be reproduced on a commercial website and are not reproduced here. The article is freely readable and the full scale can be seen there (https://doi.org/10.1007/s10803-010-1133-5). Clinicians wanting to use the RAADS-R in commercial or clinical-trial contexts should seek written permission from the authors or from the publisher.
Get our case study pack by email, with practical examples from clinical teams.
We will only use your details to send relevant Aisel resources. Unsubscribe at any time.
We value your privacy
We use cookies to analyse site usage and improve your experience. Analytics and embedded media (e.g. YouTube) only load if you accept. Read our cookie policy.