Why Social Stories work: 30 years of the Carol Gray framework, explained

If you've worked in early childhood or disability support, you have almost certainly seen a Social Story — typically a short, illustrated narrative that describes a social situation the child is about to encounter, written from the child's perspective. The format looks deceptively simple. Behind it sits a specific framework, developed by a schoolteacher named Carol Gray in the early 1990s, and a substantial research base.

What a Social Story actually is

Carol Gray first described Social Stories™ in a 1993 article in Focus on Autistic Behavior.¹ The concept was straightforward: autistic students often struggle not because they can't learn social conventions, but because the conventions are typically taught implicitly, and implicit learning is exactly the thing many autistic children find hard. A Social Story makes the implicit explicit.

Thirty-plus years later, Gray has published multiple revisions of the framework — most recently in the 15th-anniversary edition of The New Social Story Book — and the criteria have been refined into a specific, testable definition.²

Three words in that definition matter: accurately, patient, and safe. A story that sounds like a Social Story but is actually a behaviour-management script — "I will sit still. Sitting still is good" — is not a Social Story, and the research findings below do not apply to it.

The ten defining criteria (Gray, 2015)

The ten criteria that define a compliant Social Story are worth knowing because they're also the benchmark any automated Social Story tool (including ours) should meet:

1. The Story has a goal of sharing accurate information — not just managing behaviour.
2. A two-part discovery process — gathering information about the child, the situation, and the audience — precedes writing.
3. The Story has a title, introduction, body, and conclusion.
4. The Story is written for a specific audience and adapted to their receptive language, attention, and interests.
5. The Story is written in a first-person or third-person perspective, past, present, or future tense, appropriate to the audience.
6. The Story answers who, what, when, where, why, and how — factual, concrete, and literal where needed.
7. The Story follows the Social Story Formula — a specific ratio of sentence types (more on this below).
8. The Story uses clarifying, patient, and supportive language.
9. The Story is edited to refine its literal accuracy.
10. Implementation follows ten specific guidelines — including introducing the story in a calm moment, and reviewing it with a positive tone.

The seven sentence types — why the ratio matters

The most technical and most misunderstood aspect of the framework is the sentence-type ratio. Gray defines seven types of sentence a Social Story can use:

• Descriptive — accurate, factual statements about the situation.
• Perspective — statements about others' thoughts, feelings, or knowledge.
• Directive — gentle suggestions of what the child might do (rare).
• Affirmative — statements that affirm a shared value.
• Cooperative — statements about what others will do to help.
• Control — strategies the child can choose to use (older children).
• Partial — fill-in-the-blank sentences for comprehension checks.

The Social Story Formula requires that descriptive, perspective, affirmative, and cooperative sentences substantially outnumber directive sentences. The reason is crucial: stories that are heavy on "I will" and "I must" are behavioural scripts — not Social Stories — and the evidence base does not support them as effective.

What 30 years of research actually shows

There are now several systematic reviews and meta-analyses of Social Stories research. The most frequently cited are Karkhaneh et al. (2010), Test et al. (2011), and Kokina & Kern (2010).

Karkhaneh et al. (2010) — Cochrane-style systematic review

Published in Autism, this review screened the literature through May 2007 and included six randomised controlled trials and controlled clinical trials. The authors concluded that Social Stories "may be beneficial in modifying social skills in children on the autism spectrum", particularly those with at least grade-one reading skills or functional verbal/non-verbal communication.³

The review also flagged the key limitations that subsequent research has echoed: limited evidence on long-term maintenance, on effectiveness in less-controlled settings, and on optimal dose/frequency.

Test et al. (2011) — comprehensive meta-analysis

This 2011 comprehensive review and meta-analysis synthesised 28 studies and concluded that Social Stories can be considered an evidence-based practice under the criteria adopted by the National Professional Development Center on Autism Spectrum Disorder.⁴ The authors cautioned that effectiveness depends substantially on fidelity to the Gray framework — i.e. following the criteria above.

Kokina & Kern (2010) — meta-analysis of 18 studies

This meta-analysis in the Journal of Autism and Developmental Disorders examined 18 single-case design studies and found moderate effects overall. Importantly, the researchers identified that characteristics of the target child (age, reading ability, specific diagnosis) and features of the intervention (length, review frequency, implementer training) both moderated the effect.⁶

Why Social Stories are on the NPDC Evidence-Based Practice list

The National Professional Development Center on Autism Spectrum Disorder (NPDC) maintains a rigorous list of evidence-based practices for people with autism, updated periodically after systematic literature reviews. In the most recent comprehensive review (Wong et al., 2015), Social Narratives — which includes the Social Story framework — are formally classified as an evidence-based practice for children and youth.⁵

To land on the NPDC list, an intervention must be supported by specific evidence thresholds: at least two high-quality experimental or quasi-experimental group studies, or five high-quality single-case design studies, conducted by at least three different researchers across three geographic locations. Social Narratives clear that bar.

The honest limits

Three limitations we think every educator using Social Stories should know:

• The evidence is strongest for autistic children with basic verbal or reading-based comprehension. For children with severe intellectual disability, minimal language, or very young pre-verbal children, the evidence is thinner and other supports may work better.
• Effects are generally modest and often child-specific. A well-constructed Social Story does not always "work". Response varies by child, by the story's fidelity to the Gray framework, and by how it is introduced and reviewed.
• Maintenance and generalisation are not automatic. A story read once a week for three weeks and then retired is unlikely to produce lasting change. Embedding in a broader plan that includes practice, reinforcement, and co-regulation is what the evidence actually supports.

How to use Social Stories well

Drawing on the Gray framework and the reviews above, six practices we think are worth bringing to every Social Story:

Our own Social Story generator follows the Gray ten-criteria framework — we describe the mechanics on our Social Stories Explained page. We think this is one of the places where AI can genuinely help educators, provided the underlying framework is respected.