The research drew on 905 autistic children enrolled in the Shanghai Autism Early Developmental study, a cohort that recruits children diagnosed with autism spectrum disorder at Xinhua Hospital. This sample size matters enormously in psychometric research, the branch of psychology concerned with the theory and technique of psychological measurement. Small validation studies can produce unstable estimates of reliability and validity that collapse when a test is used in new populations. By evaluating the scale in nearly a thousand young children, the researchers could examine not just whether the instrument works overall, but whether it performs consistently across clinically meaningful subgroups defined by age, sex, severity of autistic symptoms, and intellectual functioning or developmental status. That subgroup analysis is what elevates the study from a routine validation exercise to a genuinely informative test of how anxiety presents in this population.

The Preschool Anxiety Scale asks parents to rate the frequency of anxiety-related behaviors in their child, covering domains such as separation anxiety, social phobia, generalized anxiety, obsessive-compulsive behaviors, and fears of physical injury. For a scale to be considered trustworthy, it must demonstrate several distinct properties. Internal consistency, typically quantified by Cronbach’s alpha, reflects how coherently the items hang together as a measure of a single underlying construct. Construct validity concerns whether the statistical structure of the items matches the theoretical organization of anxiety. Convergent validity asks whether scores correlate with other measures of related constructs, while divergent validity confirms that the scale does not simply register any and all childhood distress.

On the core reliability metric, the results were striking. The scale achieved a Cronbach’s alpha of 0.896 across the full sample, a value comfortably within the range conventionally regarded as good, and one that indicates parents’ responses to the individual items were highly coherent. Equally important, this internal consistency held up across the examined subgroups, suggesting the instrument behaves similarly whether the child is a two-year-old or a six-year-old, a boy or a girl, or a child with more or less severe autistic features. For clinicians screening a heterogeneous population, this stability is essential: a questionnaire that works only for verbal, high-IQ children would be of limited use in a clinical setting where intellectual disability and global developmental delay are common.

The validity analyses provided further support. The researchers correlated PAS scores with the internalizing and externalizing subscales of the Strengths and Difficulties Questionnaire, a widely used behavioral screening instrument. Internalizing scores capture emotional symptoms such as worry and unhappiness, while externalizing scores reflect hyperactivity, conduct problems, and impulsive behavior. If the PAS truly measures anxiety, it should correlate more strongly with internalizing than with externalizing symptoms. That is precisely what the team found: a Pearson correlation of 0.278 with the SDQ internalizing score, compared with a weaker 0.151 with the externalizing score. The pattern of a stronger association with the conceptually related construct and a weaker one with the unrelated construct satisfies the classic requirements for convergent and divergent validity, and the same pattern was reproduced within the demographic and clinical subgroups.

But the study also delivered a surprise, and it is here that the findings become most scientifically interesting. When the researchers subjected the item responses to confirmatory factor analysis, the original five-factor model of the PAS—the structure separating separation anxiety, social phobia, generalized anxiety, obsessive-compulsive behaviors, and physical injury fears—showed suboptimal fit. In other words, the neat theoretical partitioning of preschool anxiety into five distinct syndromes did not hold in this autistic sample. An exploratory factor analysis, which lets the data reveal their own structure rather than testing a predetermined one, pointed toward a far more parsimonious solution: either a single general anxiety factor or a two-factor structure. The implication is that in young autistic children, anxiety symptoms as captured by the PAS may be less clearly differentiated than in typically developing preschoolers.

There are several plausible reasons for this structural shift, and the authors’ findings invite careful interpretation. One possibility is developmental: the boundaries between anxiety subtypes may simply not yet be carved out in children this young, particularly when communication difficulties limit how children can express the content of their fears. Another possibility is specific to autism itself. Overlap between autistic traits and anxious behaviors—restricted interests that resemble compulsions, social withdrawal that resembles social phobia, distress at change that resembles generalized anxiety—could blur the distinctions between subscales, causing items from different theoretical domains to load together. A third consideration is cultural and linguistic: the scale was completed by Chinese-speaking parents, and cross-cultural differences in how anxiety behaviors are described and noticed could influence the factor structure. The study cannot disentangle these explanations definitively, but the practical message is clear: the total PAS score is a reliable and valid index of overall anxiety burden in this population, while interpretation of individual subscale scores should be made with caution until the instrument is further adapted for autistic preschoolers.

The clinical stakes of this work are considerable. Anxiety in autistic children is associated with poorer outcomes across domains ranging from sleep and feeding to family stress and school readiness, and effective interventions exist—but only if anxiety is recognized in the first place. In children aged two to six, whose autistic diagnosis has often just been established and whose families are navigating a new and overwhelming clinical landscape, a brief parent-report measure that reliably flags anxiety could change the trajectory of early intervention. The subgroup analyses conducted by the Shanghai team suggest that clinicians can deploy the PAS across their entire preschool autism caseload without worrying that the instrument will behave differently for children with global developmental delay or intellectual disability, groups that are frequently excluded from validation research and therefore poorly served by standard assessment tools.

The study also carries a broader methodological lesson for the field of autism research. Instruments developed and validated in general populations are routinely imported into autistic samples on the assumption that they measure the same construct in the same way. This study demonstrates both the value and the limits of that assumption. The reliability and convergent validity findings justify the use of the PAS total score; the factor-analytic findings warn against uncritical use of the original scoring structure. The authors themselves frame this as a call for future adaptation of the scale to this population, and their large, well-characterized cohort provides exactly the kind of data foundation that such adaptation requires. It is a model of how psychometric validation should proceed: not merely asking whether a test works, but probing where it begins to bend.

For researchers and clinicians outside China, the findings raise an obvious question of generalizability, and replication in other linguistic and cultural contexts will be needed before the conclusions can be considered universal. Still, the core result—that a brief parent-report anxiety scale can be trusted to measure overall anxiety in autistic toddlers and preschoolers, with a caveat about its internal structure—represents a meaningful step forward. Early identification of anxiety in autism has long been hampered by the absence of validated tools for the youngest children, and this study, grounded in the Shanghai Autism Early Developmental cohort and published open access in BMC Psychiatry, offers clinicians a tested instrument and offers researchers a clearer picture of how anxiety is organized in the developing autistic brain. As screening programs for autism expand worldwide, ensuring that the emotional lives of these children are measured as carefully as their diagnostic symptoms may prove one of the most consequential tasks in developmental pediatrics.

Subject of Research: Validation of the Preschool Anxiety Scale for measuring anxiety in young autistic children

Article Title: Psychometric properties of the Preschool Anxiety Scale in 2-6-year-old Chinese autistic children

Article References: Wang, M., Chen, R., Huang, C., Tan, H., Chen, J., Huang, L., Pu, Y., Zhang, L., Li, F., Du, W., Zhou, W., & Ren, T. (2026). Psychometric properties of the Preschool Anxiety Scale in 2-6-year-old Chinese autistic children. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08666-x

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08666-x

Keywords: autism spectrum disorder, anxiety, psychometrics, Preschool Anxiety Scale, preschool children, validation study, internal consistency, factor analysis, Strengths and Difficulties Questionnaire, China, early intervention, BMC Psychiatry