Reading a GARS-3 Score in Context

What the GARS-3 Is and What It Produces

The Gilliam Autism Rating Scale, Third Edition is a norm-referenced instrument of 58 items organized into six subscales, intended for individuals aged 3 through 22, and used either to screen for autism spectrum disorder or to contribute to a fuller autism evaluation. Results come back as scaled scores on each subscale plus an overall Autism Index, and the scoring runs in the symptomatic direction: higher means greater likelihood of autism 1Ref 1Karren BC (2017).A Test Review: Gilliam, J. E. (2014). Gilliam Autism Rating Scale–Third Edition (GARS-3).That the GARS-3 comprises 58 items across six subscales, is norm-referenced for ages 3 to 22, is used to screen for or contribute to an autism evaluation, and reports subscale scaled scores plus an Autism Index in which higher indicates greater likelihood of ASD; that this peer-reviewed review is the citable source of record standing in for the commercial manual; and the review's stated cautions regarding ages 20 to 22, minority populations, and a norming sample composed only of individuals with ASD diagnoses with limited racial and ethnic diversity. Deliberately not cited for cutoffs or accuracy figures..

Its publisher is a commercial test company and the test manual is not a peer-reviewed document, so the citable source of record for what this instrument is and how it behaves is an independent test review published in the Journal of Psychoeducational Assessment 1Ref 1Karren BC (2017).A Test Review: Gilliam, J. E. (2014). Gilliam Autism Rating Scale–Third Edition (GARS-3).That the GARS-3 comprises 58 items across six subscales, is norm-referenced for ages 3 to 22, is used to screen for or contribute to an autism evaluation, and reports subscale scaled scores plus an Autism Index in which higher indicates greater likelihood of ASD; that this peer-reviewed review is the citable source of record standing in for the commercial manual; and the review's stated cautions regarding ages 20 to 22, minority populations, and a norming sample composed only of individuals with ASD diagnoses with limited racial and ethnic diversity. Deliberately not cited for cutoffs or accuracy figures.. That sounds like bibliographic housekeeping. It is not. It is the reason several numbers attached to this test online have no independent backing at all.

The subscale profile carries more usable information than the single Index number. Six subscales exist because the behaviors they cover do not move in lockstep, and one person can land in very different places across them. The pattern is what an evaluator reads. The Index is what gets quoted.

What an Autism Index Actually Compares You To

Norm-referenced means a score is not a count of anything. It is a position — where this person's ratings fall relative to the sample of people the test was standardized on. Change the reference sample and an identical set of ratings produces a different number. That is true of every norm-referenced test ever built, and it is the most useful single thing to understand about an Index score.

So the makeup of the norming sample is not a footnote. It is part of the measurement. The peer-reviewed review of the GARS-3 states that its norming sample consisted only of individuals who already carried autism diagnoses, and that it had limited racial and ethnic diversity 1Ref 1Karren BC (2017).A Test Review: Gilliam, J. E. (2014). Gilliam Autism Rating Scale–Third Edition (GARS-3).That the GARS-3 comprises 58 items across six subscales, is norm-referenced for ages 3 to 22, is used to screen for or contribute to an autism evaluation, and reports subscale scaled scores plus an Autism Index in which higher indicates greater likelihood of ASD; that this peer-reviewed review is the citable source of record standing in for the commercial manual; and the review's stated cautions regarding ages 20 to 22, minority populations, and a norming sample composed only of individuals with ASD diagnoses with limited racial and ethnic diversity. Deliberately not cited for cutoffs or accuracy figures.. A score therefore answers a narrower question than it appears to be answering.

The practical translation is this. An Index tells you how someone's rated behavior compares with a particular reference group under particular conditions. It does not tell you how autistic someone is, because that is not a quantity that exists to be measured. And no rating instrument is the gold-standard autism assessment — that role belongs to a full diagnostic evaluation by clinicians who have met the person.

The Cautions Its Peer Review Attaches

The review of record recommends caution in two specific situations: assessing individuals aged 20 to 22, and assessing people from minority populations 1Ref 1Karren BC (2017).A Test Review: Gilliam, J. E. (2014). Gilliam Autism Rating Scale–Third Edition (GARS-3).That the GARS-3 comprises 58 items across six subscales, is norm-referenced for ages 3 to 22, is used to screen for or contribute to an autism evaluation, and reports subscale scaled scores plus an Autism Index in which higher indicates greater likelihood of ASD; that this peer-reviewed review is the citable source of record standing in for the commercial manual; and the review's stated cautions regarding ages 20 to 22, minority populations, and a norming sample composed only of individuals with ASD diagnoses with limited racial and ethnic diversity. Deliberately not cited for cutoffs or accuracy figures.. Both cautions trace back to the norming sample rather than to anything about the person being rated. An instrument generalizes only as far as the group it was built on reaches, and the review says plainly where this one thins out.

This page prints no GARS-3 cutoff, and no sensitivity or specificity figure. The independent review establishes the instrument's structure and its psychometric cautions 1Ref 1Karren BC (2017).A Test Review: Gilliam, J. E. (2014). Gilliam Autism Rating Scale–Third Edition (GARS-3).That the GARS-3 comprises 58 items across six subscales, is norm-referenced for ages 3 to 22, is used to screen for or contribute to an autism evaluation, and reports subscale scaled scores plus an Autism Index in which higher indicates greater likelihood of ASD; that this peer-reviewed review is the citable source of record standing in for the commercial manual; and the review's stated cautions regarding ages 20 to 22, minority populations, and a norming sample composed only of individuals with ASD diagnoses with limited racial and ethnic diversity. Deliberately not cited for cutoffs or accuracy figures.; the accuracy figures that circulate for this test come from publisher materials rather than from that review, and the gap between a vendor's number and an independently reported one is invisible to a reader once the number has been copied onto a third-party page.

If a report in front of you carries an Index score, the questions worth asking are who completed the ratings, what the subscale profile looked like underneath the composite, and what the evaluator concluded after setting it beside everything else they had. Being unable to interpret a standardized score without the manual is not a gap in your understanding. It is how these instruments are designed.

Where the GARS-3 Sits Among the Other Instruments

Autism instruments differ in a way that matters more than their names: who supplies the information, and whether anyone interacts with the child. The GARS-3 gathers ratings about a person from someone who knows them 1Ref 1Karren BC (2017).A Test Review: Gilliam, J. E. (2014). Gilliam Autism Rating Scale–Third Edition (GARS-3).That the GARS-3 comprises 58 items across six subscales, is norm-referenced for ages 3 to 22, is used to screen for or contribute to an autism evaluation, and reports subscale scaled scores plus an Autism Index in which higher indicates greater likelihood of ASD; that this peer-reviewed review is the citable source of record standing in for the commercial manual; and the review's stated cautions regarding ages 20 to 22, minority populations, and a norming sample composed only of individuals with ASD diagnoses with limited racial and ethnic diversity. Deliberately not cited for cutoffs or accuracy figures.. A free parent-report screen works differently again — the M-CHAT-R/F asks caregivers 20 questions about toddlers aged 16 to 30 months, with a structured follow-up interview when a screen comes back positive, and it is a screen rather than a diagnostic test 2Ref 2Robins DL, Fein D, Barton M (2009).M-CHAT-R/F (Modified Checklist for Autism in Toddlers, Revised, with Follow-Up) — official screening instrument.That the M-CHAT-R/F is a free 20-item parent-report screen for toddlers aged 16 to 30 months with a structured two-stage follow-up interview for positive screens, and that it is a screening instrument rather than a diagnostic test. No items are reproduced or paraphrased here..

Some instruments are done with the child directly. The STAT is an interactive, play-based screener for children aged 24 to 36 months that works through play, communication, and imitation skills. Its cutoff was derived through signal detection analysis, its sensitivity, specificity, and predictive values were confirmed in both a development sample and an independent validation sample, and its risk categories were checked against ADOS-G classification 3Ref 3Stone WL, Coonrod EE, Turner LM, et al. (2004).Psychometric Properties of the STAT for Early Autism Screening.That the STAT is an interactive, play-based Level 2 autism screener for children aged 24 to 36 months assessing play, communication, and imitation; that its cutoff was derived through signal detection analysis with sensitivity, specificity, and predictive values confirmed in a development and an independent validation sample; and that its risk categories were compared against ADOS-G classification. No numeric cutoff, item count, or score range is asserted..

Others are rated by an observer who watched. The CARS-2 rating scale is that kind: fourteen areas of behavior plus a general-impressions rating, each scored 1 to 4 in half-point steps, giving a total between 15 and 60 in which higher means greater impairment 4Ref 4Schopler E, Reichler RJ, DeVellis RF, Daly K (1980).Toward objective classification of childhood autism: Childhood Autism Rating Scale (CARS).The observer-rated structure the CARS-2 descends from: fourteen behavioral areas plus a general-impressions rating, each scored 1 to 4 in half-point increments, giving a 15-to-60 total in which higher indicates greater impairment. Used here only for the contrast in instrument type; no cutoff bands are cited..

None of these is a smaller or larger version of another. They collect different evidence, from different people, at different ages. A full evaluation is where several of them get reconciled against each other.

Why Who Filled It In Matters as Much as What It Says

A rating instrument records one person's account of another person's behavior, which means the account arrives carrying the rater's vantage point. An adult who sees a child only at home sees a different child from one who sees them only in a classroom of twenty-four. Neither is wrong. Both are partial, and one set of ratings from one rater is one vantage point.

That is why teacher input is requested so often during autism evaluations, and why evaluators put weight on cross-setting observation. Agreement between raters in different settings is informative; disagreement is informative too. A child who reads as unremarkable at school and comes apart at home every afternoon is telling you something specific about where their effort is being spent.

Diagnosis is not a matter of accumulating rating forms. It rests on developmental history and directly observed behavior, and a comprehensive evaluation may draw on developmental pediatricians, child psychologists or psychiatrists, and neurologists, because no laboratory test settles the question 5Ref 5Centers for Disease Control and Prevention (2024).Clinical Testing and Diagnosis for Autism Spectrum Disorder.That autism diagnosis rests on developmental history and observed behavior rather than a laboratory test, and that a comprehensive evaluation may involve developmental pediatricians, child psychologists or psychiatrists, and neurologists.. Rating scales feed that process rather than shortcut it.

What Happens at the Top of the Age Range

The GARS-3 stops at 22, and its review recommends particular caution in the 20-to-22 band 1Ref 1Karren BC (2017).A Test Review: Gilliam, J. E. (2014). Gilliam Autism Rating Scale–Third Edition (GARS-3).That the GARS-3 comprises 58 items across six subscales, is norm-referenced for ages 3 to 22, is used to screen for or contribute to an autism evaluation, and reports subscale scaled scores plus an Autism Index in which higher indicates greater likelihood of ASD; that this peer-reviewed review is the citable source of record standing in for the commercial manual; and the review's stated cautions regarding ages 20 to 22, minority populations, and a norming sample composed only of individuals with ASD diagnoses with limited racial and ethnic diversity. Deliberately not cited for cutoffs or accuracy figures. — which is precisely the age at which many people first go looking for an answer about themselves. Adults never assessed as children are a large group, and instruments built and normed on children do not simply extend upward into adult life.

Adult assessment uses adult instruments. The RAADS-R is one: an 80-question scale developed to assist the diagnosis of autism spectrum disorder in adults, organized across social relatedness, circumscribed interests, language, and sensory-motor domains. Its international validation ran across 779 people — 201 with autism and 578 comparison subjects — and reported sensitivity of 97%, specificity of 100%, test-retest reliability of .987, and concurrent validity of 95.59% against the Social Responsiveness Scale-Adult 6Ref 6Ritvo 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.That the RAADS-R is an 80-question scale assisting adult autism diagnosis across social relatedness, circumscribed interests, language, and sensory-motor domains; its international validation across 779 subjects (201 ASD, 578 comparisons) whose ASD participants already met established diagnostic criteria; and its reported sensitivity of 97%, specificity of 100%, test-retest reliability of .987, and concurrent validity of 95.59% with the Social Responsiveness Scale-Adult. No score range or cutoff is cited..

Those are strong figures, and they come from a study in which the autism group had already been diagnosed against established criteria before the scale was applied 6Ref 6Ritvo 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.That the RAADS-R is an 80-question scale assisting adult autism diagnosis across social relatedness, circumscribed interests, language, and sensory-motor domains; its international validation across 779 subjects (201 ASD, 578 comparisons) whose ASD participants already met established diagnostic criteria; and its reported sensitivity of 97%, specificity of 100%, test-retest reliability of .987, and concurrent validity of 95.59% with the Social Responsiveness Scale-Adult. No score range or cutoff is cited.. That is the normal shape of an instrument validation, and it describes how cleanly the scale separated groups that were already separated. It is a different claim from how any questionnaire performs on an unsorted person walking into a clinic for the first time.

What to Do With a GARS-3 Result

Treat it as one document in a file rather than as an answer. A GARS-3 result contributes to an evaluation; the diagnosis comes from a clinician integrating developmental history, direct observation, and testing, sometimes across several specialties 5Ref 5Centers for Disease Control and Prevention (2024).Clinical Testing and Diagnosis for Autism Spectrum Disorder.That autism diagnosis rests on developmental history and observed behavior rather than a laboratory test, and that a comprehensive evaluation may involve developmental pediatricians, child psychologists or psychiatrists, and neurologists.. If a rating result is the only thing a family has, what they have is a reason to pursue an evaluation, not the outcome of one.

The useful requests are specific. Ask what the subscale profile showed rather than only the Index. Ask who completed the ratings and how much of the person's life that rater actually sees. Ask what the evaluator would expect to find if this particular result turned out to be misleading — clinicians who work with these instruments have a ready answer, and the answer tells you how much weight they are putting on it themselves.

When cost or waiting is what stands between a rating result and a real evaluation, two routes sometimes open things up: a university training clinic, where supervised trainees evaluate at reduced fees, and a telehealth evaluation, which widens the geography a family can choose from. Both carry limits worth asking about directly. Both are better than another year holding a questionnaire score nobody has interpreted.