Utility of the Gilliam Autism Rating Scale in research and clinical populations.

South et al. (2002) checked if the Gilliam Autism Rating Scale (GARS) spots autism as well as doctors do. They compared GARS scores to DSM-IV diagnoses in a mixed-age sample of children with autism.

The study asked: does a quick parent checklist catch every child a clinician would diagnose? The answer shapes how much weight teams give the GARS during intake.

Sensitivity was only 0.48. That means the GARS missed about half of the children who actually met DSM-IV criteria for autism.

The scale often labeled autistic kids as non-autistic, so teams who rely on GARS alone risk delaying services.

Goodwin et al. (2012) saw the same problem in adults. Their Autism-Spectrum Quotient also missed the majority of people with ASD, showing the GARS flaw is not age-specific.

Mohanakumar Sindhu et al. (2025) echoed the warning with language screeners. CELF-5 and Vineland-3 likewise showed poor sensitivity for autistic kids, proving that quick checklists can miss multiple domains.

So et al. (2024) offers a bright contrast. A robot observer reached 0.88 or higher sensitivity by tracking eye contact and attention. Their high-tech tool outperformed the paper-and-pencil GARS, hinting that newer methods may replace old questionnaires.

If you screen with GARS, treat a low score as "not sure," not "no autism." Always add a clinician interview, ADOS, or other well-validated tool. The same rule applies to any brief scale: parent and self-report checklists are helpful alerts, not final verdicts. Build a multimodal battery so kids don’t slip through the cracks while waiting for answers.