Early identification of social communication needs: Evidence of a cross-cultural preschool assessment

Young teacher and toddler playing with plastic food and cutlery toys at kindergarten

Penny Roy, Shula Chiat & Jennifer Warwick

This article was originally published in Assessment & Development Matters Vol. 12 No. 3 Autumn 2020

Background and rationale for development of the Early Sociocognitive Battery (ESB)

Two lines of evidence have shown that the needs of children with ‘late’ diagnosed autism spectrum disorders (ASD) can go unrecognised for years. First, pre-schoolers referred to Speech and Language Therapy (SLT) services with concerns about their language and communication are at significant risk of social communication difficulties in the longer-term and ‘late’ diagnosis of autism spectrum disorders (ASD), often several years after their initial referral. Second, studies of children with ASD show that a substantial proportion are only diagnosed at 5–6 years of age or later. For at least some of these children, earlier language delay/disorder can be a factor. These children and their families miss opportunities to benefit from the support, understanding and appropriately targeted interventions afforded by early identification.

What is the ESB?

The ESB was developed as a research tool to measure early skills of social engagement and understanding that are known to be associated with early language developmentand impaired in children with ASD. It provides systematic measurement of skills typicallyobserved informally by therapists when they assess young children, for example, how theyuse eye contact, share interest in a toy or activity with another or engage in pretend play.

It comprises three subscales:

  • Social Responsiveness: responsiveness to an adult’s expression of feelings
  • Joint Attention: capacity to share attention with another on the same object
  • Symbolic Comprehension: understanding of symbols (including gestures, miniature objects and substitute/‘pretend’ objects)

The tasks involve minimal verbal comprehension and responses are non-verbal, elicited through evidence-based, carefully selected, and well-specified activities. The assessment as a whole can be administered in about 15 minutes, compliance is high, and as the aim is to measure social engagement, non-compliance is informative (Roy & Chiat, 2019).

The test yields raw, categorical, and scaled/standard scores. Online scoring and categorisation of subscale and total raw scores as ‘Low’ are useful for initial profiling of skills, identification of risk and more immediate feedback; reference to the norms and conversion of raw scores to scaled and standard scores allow for a more nuanced profiling of children’s strengths and difficulties.

UK standardisation of the ESB and cross-cultural evidence

Standardisation of the ESB enables clinicians to make objective comparisons of a child’s response with those typical for their age. The final UK standardisation sample comprised 384 children across five age bands (four 6-month age bands from 2:0 to 3:11, and a one-year age band 4:0-4:11), excluding children with ongoing contact with SLT services (N=32). Parents completed background questionnaires and the distribution of gender, ethnicity, English as an additional language and primary carer’s educational qualifications was in line with current UK social metrics. The ESB and the British Picture Vocabulary Scale-3rd edition (BPVS3), a measure of receptive vocabulary, were administered by ESB-trained SLTs. Table 1 shows the effects of child/demographic factors on children’s performance.

Table 1 shows the effects of child/demographic factors on children’s performance.

Table 1

Note: p values: ns non-significant, **<.01, ***<.001; +taking age as a covariate; PC: Primary carer; EAL: English as an additional language. η2 partial eta squared. n=416 (including children with SLT contact)

As can be seen in Table 1, ESB performance was unaffected by gender or ethnicity and both measures were sensitive to age and contact with SLT services (indicative of language delay). The findings that scores on the ESB, in contrast to the BPVS3, were unaffected by bi/multilingualism and less affected by socioeconomic status (as measured by PC educational qualifications) show that the ESB is suitable for use with children from diverse language and cultural backgrounds.

Further cross-cultural evidence

The above evidence from the standardisation was restricted to EAL and ethnic groups of children currently living in the UK. A study using the ESB with young children living in Riyadh, Saudi Arabia, conducted by a PhD student from City, University of London (Alkadhi, 2015) provides additional cross-cultural evidence. In the absence of standardised tests of early language delay, Dr Alkadhi developed a battery of assessments for Saudi Arabic-speaking children and administered these, together with the ESB, to 161 children aged 2;0-3;5 years. Table 2 compares the mean Total ESB scores from the UK standardisation sample with their age-matched peers in Saudi Arabia.

Table 2 compares the mean Total ESB scores from the UK standardisation sample with their age-matched peers in Saudi Arabia.

Table 2

Note: *Cohen’s d effect sizes

As can be seen in Table 2, the mean ESB scores in the two samples were closely comparable across age groups supporting the cross-cultural validity of the ESB.

Psychometric properties: Reliability and validity

As estimates of reliability and validity are only meaningful when applied to specificpopulations from which data were derived, evidence of the psychometric robustness ofthe ESB is drawn from three different samples: the UK normative sample and the sampleof Saudi Arabian preschool children (see above), and a follow-up sample of UK preschoolchildren referred to SLT services with concerns about language and communication(N=209 when first seen). Estimates of inter-rater and test-re-test reliability and construct,concurrent and predictive validity support the psychometric robustness of the ESB, withcoefficients on key measures of reliability ranging from .8-≥.9.

Predictive validity

 Figure 1 shows a scattergram of preschool Total ESB scores (y axis) and SRS Total T-scores (x axis) for the youngest age group (2;6-2;11)

Figure 1

The predictive validity of the ESB was derived from outcomes in the follow-up sampleof clinically referred children (aged 2;6-3;11) when they were in their last year ofprimary school education (aged 9–11 years). No child had a clinical diagnosis of ASDwhen first seen as pre-schoolers. At the final follow-up, 7–8 years later, just under athird had social communication difficulties (SCD) according to parent ratings on theSocial Responsiveness Scale (SRS), and nearly one-fifth (19.4 per cent) had a parent-reportedclinical ‘late’ diagnosis of SCD and/or ASD. Compared with early preschoolmeasures of language and behavioural problems, the ESB was a better predictor ofsocial communication difficulties and ongoing contact with SLT services. Using cut-offsfor ‘Low’ scores from the standardisation sample, ‘Low’ ESB identified 89 per cent ofchildren with ‘late’ SCD&/ASD diagnoses (sensitivity) and 75 per cent of those without(specificity). Low ESB scores were equally predictive of parent-reported clinical diagnosisacross age groups. Figure 1 shows a scattergram of preschool Total ESB scores (y axis) andSRS Total T-scores (x axis) for the youngest age group (2;6-2;11). The dotted Y referenceline represents the cut-off for Low ESB scores (-2SDs), and the two X reference linescut-off scores of 60 and 76 on the SRS. Scores ≥76 are in the severe range and are stronglyassociated with clinically diagnosed ASD, and those in the 60-75 range are classified as‘mild to moderate’. The set markers (see legend) identify children who had or had notreceived a ‘late’ clinical diagnosis of SCD&/ASD according to parental report.

As can be seen above in Figure 1, the correlation between Total ESB scores and SRS Total T-scores, 7-8 years later, was strong (r=-.71). All children with SRS scores in the severe range had Low ESB scores as pre-schoolers, as was the case, with the exception of one child, for those with a ‘late’ clinical diagnosis of SCD &/ASD. See Roy, Chiat and Warwick (2019) for full standardisation and psychometric evidence.


The Early Sociocognitive Battery is a standardised, evidence-based clinical assessment for children aged 2;0-4;11. It is a psychometrically robust, preschool measure suitable for use with children from diverse language and cultural backgrounds. Low ESB performance identifies preschool children who are at risk of short- and long-term social communication difficulties and ASD as well as early language difficulties. It provides profiles of pivotal sociocognitive skills that can guide early and appropriately targeted intervention to support these children and monitor change across time. It is also suitable for older children diagnosed with ASD as a measure of preverbal social communication (Taylor et al., 2020).

The authors

Penny Roy is an Honorary Professor of Developmental Psychology, City, University of London. Shula Chiat is an Honorary Professor of Child Language, City, University of London. Jennifer Warwick is a Specialist Speech and Language Therapist.


Roy, P. & Chiat, S. (2019). The Early Sociocognitive Battery: A clinical tool for early identification of children at risk for social communication difficulties and ASD? International Journal of Language & Communication Disorders, 54, 794–805.

Roy, P., Chiat, S. & Warwick, J. (2019). Early Sociocognitive Battery ESB. Oxford: Hogrefe Ltd.

Taylor, L.T., Charman, T., Howlin, P., Slonims, V., Green, J. & The PACT-G Consortium (2020). Brief report: Associations between preverbal social communication skills, language, and symptom severity in children with autism: An investigation using the Early Sociocognitive Battery. Journal of Autism and Developmental Disorders, 50, 1434–1442.