Social Stories for Autistic Children: What the Latest Research Really Shows

The short answer

A Social Story is a short, personalised way to share clear information about one situation. It might explain what happens during a fire alarm, how a child can ask for a break, or what to expect when a familiar teacher is away.

It is not a treatment or cure for autism. The NHS explains that autism is not an illness. Autistic children may need support, but they do not need to be trained to look less autistic.

The honest verdict is that a well-made Social Story may help with one clear, meaningful situation. It should not be sold as a general autism therapy. The child should be involved, comfortable and free to stop. Adults may need to change the environment too.

What is a genuine Social Story?

The approach was developed by Carol Gray. A genuine Social Story follows a defined process and ten criteria. It is researched and written for one person. It should be accurate, descriptive, respectful and safe.

This matters because many online resources labelled “social stories” are really instruction sheets. A list telling a child to sit still, make eye contact and “behave well” is not the same thing.

A useful story about a fire-alarm practice could explain:

• when the alarm is likely to sound
• why it may be loud
• where the child can go
• how they can protect their ears or ask for help
• what adults will do
• what is likely to happen next

The aim is to share useful information, not to demand obedience. Carol Gray's current Social Stories 10.4 criteria also recognise that a different solution may sometimes be better than a story.

What did the large UK trial test?

The ASSSIST-2 study was a cluster randomised controlled trial. In plain English, 87 schools were placed by chance into one of two groups: Social Stories plus usual support, or usual support alone.

The trial included 249 children aged 4 to 11 in Yorkshire and the Humber whose parent or carer reported a clinical autism diagnosis. Intended eligibility also required a teacher to report daily behaviour described by the study as challenging, so the findings may not apply to every autistic child. Mainstream and special schools took part. In the Social Stories group:

1. A behavioural goal was selected at a short meeting with the teacher, parent or carer and a researcher, sometimes with a teaching assistant. Children contributed only occasionally. If adults disagreed, the teacher's preferred school-based goal was used.
2. School interventionists received training. Parents and carers were invited to attend.
3. A trained member of staff delivered a personalised story at school.
4. The story was meant to be read with the child at least six times over four weeks.
5. Researchers checked outcomes after six weeks and six months.

The stories were added to usual support. They did not replace it. The planned trial intervention was delivered at school, so its results should not be treated as proof for a copied template or an informal story used at home.

What did the trial find?

No clear change in broad social responsiveness

The main outcome was the teacher-rated Social Responsiveness Scale-2 T-score at six months. Lower scores mean fewer reported difficulties. The teacher completing this measure was not the staff member delivering the story, but still knew which group the school was in.

The adjusted intervention-minus-usual-care difference was −1.61 points. The 95% confidence interval ran from −4.18 to 0.96, and p=0.220. The interval spans no difference, a benefit of up to 4.18 points and a worsening of up to 0.96 points, so the trial did not detect a statistically clear improvement in this broad teacher-rated outcome.

Show the trial number

The point estimate was smaller than the 3-point difference used to plan the trial. However, the confidence interval still included benefits larger than 3 points, as well as no effect and a slight worsening, so the size of any true effect remains uncertain.

A weak positive signal on one teacher-rated goal

Teachers gave higher scores for how often a pre-set behavioural goal was met in the Social Stories group. On the study's 0-to-10 rating, the estimated difference was 0.97 points, with a 95% confidence interval from 0.21 to 1.73 and p=0.012.

This is a useful signal, not objective proof of progress. The goal measure was made for each child, it was a secondary outcome, and teachers both knew which children received the story and supplied the ratings. The method also treated every step on the 0-to-10 scale as equal, an assumption the trial team called doubtful. The authors described the result as weak evidence because expectations could have influenced the scores.

No clear benefit for anxiety or other wider outcomes

The trial found no clear differences in anxiety, depression, general health or parent stress. Parents should not be told that Social Stories are proven to reduce anxiety in general.

A story may still make one event more predictable. That is different from treating an anxiety disorder.

Does reading the story six times change the result?

A secondary analysis tried to estimate the effect for children who would receive at least six sessions if offered them. At six months, the intervention-minus-usual-care estimate was −3.37 points on the broad social score (95% confidence interval −6.65 to −0.10; p=0.043); lower scores favoured Social Stories.

This was a complier-average causal effect analysis, not a simple randomised comparison of “six readings versus fewer”. Session records were missing for many children. The trial team warned that the remaining intervention and control groups might no longer be comparable and that selection bias probably explained part of the estimate. It does not prove that six readings are an effective dose.

What does the new 2026 review add?

In June 2026, researchers published a meta-analysis of 21 single-case studies involving 61 autistic people. These studies follow one person, or a few people, closely before and during an intervention.

The review reported an overall Tau-U effect of 0.743, which the authors labelled moderate, on the particular skills or behaviours being measured. It did not detect a statistically clear difference between digital and paper formats (p=0.410). That does not prove the formats are equal; the comparison was small and grouped together different digital approaches.

This sounds more positive than the UK trial, but the studies asked different questions:

• The UK trial tested broad outcomes across 249 children.
• The smaller studies usually tested one narrow, observable target in one person at a time.

Taken together, the evidence leaves room for a careful, low-pressure trial for one narrow goal, with simple monitoring. It does not support broad claims that Social Stories improve autism, general social ability or mental health.

Limits, discomfort and safety

Social Stories do not involve a medicine or procedure. That does not make every story automatically harmless.

Six adverse events were reported during the UK trial: two serious and four non-serious events across all 249 children. The two serious events were judged unrelated or unlikely to be related to Social Stories. One non-serious event was judged probably related: while a sentence about anger was read, the child covered their ears and briefly banged their head against a wall. A teaching assistant then spoke with the child.

The study recorded events reported by participants, and the trial team said six reports were far too few to establish how often harm occurs. The related event does not show that Social Stories commonly cause harm. It does show why careful wording, active observation and stopping when a child is distressed matter.

In interviews and focus groups, adults reported that some children lost interest, refused repetition or regarded a story as too long, simplistic or patronising. The children were not directly interviewed in this process evaluation. These adult reports do not tell us how common the problem is, but they show why a child's response matters.

A child covering their ears, leaving, becoming distressed, saying “no” or losing interest is giving useful information. Do not push through it. Pause, ask what is wrong, change the story or choose a different support.

The trial also had important limits:

• COVID-19 disrupted school delivery and follow-up.
• Staff and families knew which group each child was in.
• The positive individual-goal result was rated by teachers who knew about the story.
• It took place in one English region and involved children aged 4 to 11.
• The six-month primary score was available for 211 of 249 children. Whole-school loss at that point was higher in the intervention group than in usual care, although sensitivity checks did not change the main conclusion.
• Twenty-six children were later found not to meet one or more protocol eligibility criteria, but correctly remained in the main intention-to-treat analysis.
• Twenty-seven of 249 children—about 11%—were reported in non-White ethnic groups.

For transparency, the trial's lead author disclosed that he co-authored a Social Stories guide used in the study. The report says all royalties were donated to a children's hospice. Trial-unit author Catherine Hewitt also disclosed NIHR Health Technology Assessment committee roles. These disclosures do not invalidate the study, but they help readers judge its context.

When might a Social Story be worth trying?

As a practical, child-centred option—not a benefit proven by the trial—it may be reasonable when all of these are true:

• there is one clear situation the child wants or needs help to understand
• the goal matters to the child, not only to adults
• the writer knows the child and the situation well
• the words, pictures and length match the child's communication style
• the child is willing to take part
• adults will also make reasonable changes where needed
• everyone agrees how they will tell whether it helped

Be cautious if a story or paid service promises broad social gains, uses a generic template, ignores pain or sensory overload, or aims mainly for eye contact, “quiet hands” or unquestioning compliance.

Seven steps for respectful use

1. Understand the situation first

Ask what the child may be experiencing. Is the problem uncertainty, noise, pain, unclear language, bullying, an unrealistic demand or something else?

A story cannot fix an unsafe or unsuitable environment. If lunch is unbearable because the hall is too loud, quieter access may help more than a story about tolerating noise.

2. Choose one small, useful goal

Avoid “be more social” or “cope better”. Choose something the child can recognise and adults can observe. Good goals focus on understanding, access, choice or communication.

3. Involve the child

In ASSSIST-2, children contributed to goal-setting only occasionally. As a child-centred safeguard—not a result tested by the trial—ask what they want to know. Let them choose pictures, words or format where possible. A child can take part through speech, writing, pictures, AAC, gesture or another reliable form of communication.

4. Describe more than you direct

Use clear, literal language. Explain what other people will do as well as what the child may choose to do. Avoid shame, threats, false promises and rigid words such as “always” when they are not true.

5. Personalise the format

Some children prefer photographs. Others prefer drawings, plain text, audio or a short digital version. The 2026 review did not detect a clear digital-versus-paper difference, but it was too small to prove the formats are equal. Use the format the child understands and likes.

6. Watch the child's response

Adults in the process evaluation reported that some children enjoyed repetition, while others lost interest or refused. There is no good reason to keep repeating a story that the child dislikes.

7. Change the adult response too

Communication is shared. Adults should use clear language, allow processing time, accept different ways of communicating and make sensory adjustments. The National Autistic Society's communication guidance explains this well.

A Social Story should never place the whole burden on the child.

How can we tell whether it helped?

Agree one simple measure before starting. For example:

• Can the child tell or show what will happen?
• Can they choose a support, such as headphones or a break card?
• How distressed do they feel before and after the event?
• What does the child say or show about the story?

Check after a short, agreed period. Improvement should be meaningful to the child, not only convenient for adults.

If nothing changes, that is useful information. Stop, review the situation and try a different support. Do not blame the child or the parent.

Questions to ask your child's school

1. What exact situation is the story meant to help with?
2. How was my child involved in choosing the goal and checking the story?
3. Who wrote it, and what current Social Stories guidance did they use?
4. How will it match my child's language, sensory and communication needs?
5. What will adults or the environment change too?
6. How will we know whether it helped, and when will we stop or rewrite it?
7. What will happen if my child does not want to read it?

If the story is part of wider school support, our guide to SEN Support in mainstream schools explains the assess–plan–do–review cycle and how support should be recorded.

Can parents write one at home?

Parents can learn the Social Stories process. But ASSSIST-2 tested school-based delivery by trained educational staff. Parents could attend training, but they were asked not to deliver the trial story at home. Any other home stories were outside the trial, and no home-use outcomes were collected.

The 2026 review included seven participants across three parent-implemented studies. Of those, only four participants were in the two parent-implemented studies conducted at home. We therefore do not know how effective informal parent-written Social Stories are at home.

Start with Carol Gray's current criteria. If the situation is complex, ask a SENCO, autism specialist teacher or speech and language therapist who understands the child's communication. A job title alone is not enough; the person must listen to the child and understand the situation.

You do not need an expensive treatment package to ask whether one respectful, personalised story is useful.

The bottom line

The best story shares information, respects the child and helps adults listen more carefully. If it becomes a script for compliance, it has missed the point.

NICE's current under-19 autism recommendations do not specifically recommend or reject Social Stories. NICE recommends meaningful visual and environmental adjustments and a separate play-based social-communication intervention delivered by a trained professional. NICE has announced a planned update of its psychosocial recommendations.

If a new intervention is being sold with a dramatic claim, our guide to reading autism treatment headlines has a five-question evidence checklist.

---

Frequently asked questions

Do Social Stories treat autism?

No. Autism is a lifelong neurodevelopmental difference, not an illness to cure. Social Stories share information about a particular situation.

Can a Social Story reduce anxiety?

It may make one event more predictable for some children. The large UK trial did not find a clear overall reduction in anxiety, so wider claims are not justified.

What age has the best evidence?

The main UK group trial involved children aged 4 to 11, but that does not mean this is the age when Social Stories work best. The 2026 review included people aged 3 to 32, but its age-group differences were not statistically clear (p=0.930). An optimal age is unknown.

What if my child does not read?

A story can use photographs, symbols, audio or another accessible format. Match it to the child's reliable way of understanding and communicating. If you are unsure, seek advice from someone who knows the child and understands accessible communication.

Are digital Social Stories better?

The 2026 review did not detect a statistically clear difference between digital and paper formats. That does not prove they are equal. Choose what the child prefers and can use comfortably.

Does a professional need to write it?

It is not a medical prescription, but genuine Social Stories follow defined criteria. The UK trial tested a package of staff training, collaborative writing, research-team help and checks on how stories were written. It cannot tell us which part mattered or whether the findings apply to unsupported, untrained use.

---

Sources and further reading

• Wright B et al. (2025) — ASSSIST-2 pragmatic cluster randomised controlled trial, Child and Adolescent Mental Health
• Wright B et al. (2024) — full NIHR Health Technology Assessment report and economic evaluation
• Guo T et al. (2026) — meta-analysis of 21 single-case experimental studies, Frontiers in Psychology
• Carol Gray — Social Stories overview and current definition
• Carol Gray — Social Stories 10.4 criteria update
• NHS — What is autism?
• National Autistic Society — Tips for effective communication with autistic pupils
• Newcastle Hospitals NHS Foundation Trust — Understanding and supporting social interaction
• NICE CG170 — Autism spectrum disorder in under 19s: support and management

---

Disclaimer: This guide provides general information, not medical advice. It does not assess an individual child or replace advice from professionals who know them. If a child is distressed, in pain, at risk or showing a sudden change in behaviour, seek appropriate medical, mental-health or safeguarding help.