Adult Autism Spectrum Quotient (AQ-10) – Self-administered

1. Abstract

The Adult Autism Spectrum Quotient (AQ-10) – Self-administered is a brief, 10-item clinical screening instrument derived from the original 50-item Autism-Spectrum Quotient (AQ) developed by Simon Baron-Cohen and colleagues. Designed by Carrie Allison, Bonnie Auyeung, and Simon Baron-Cohen in 2012 at the Autism Research Centre (University of Cambridge), the AQ-10 was constructed to address the pressing clinical need for a rapid, reliable, and valid triage measure in primary care and general adult psychiatric settings. The instrument samples two representative items from each of the five core psychological domains identified in the full AQ: Social Skill, Attention Switching, Attention to Detail, Communication, and Imagination. Respondents rate statements using a 4-point response scale ranging from “Definitely Agree” to “Definitely Disagree,” which are subsequently collapsed into a binary scoring framework (0 or 1 point per item) that yields a total score between 0 and 10.

Psychometric evaluations reported in the seminal development study demonstrated high sensitivity (0.88), specificity (0.91), and positive predictive value (0.85) at a clinical cutoff score of 6 or greater (>6, i.e., 7 or above, or interpreted as ≥6 depending on local implementation protocols), meeting the criteria established by the National Institute for Health and Care Excellence (NICE Clinical Guideline CG142) for adult autism identification. While initially demonstrating robust diagnostic classification properties in clinical referral samples, subsequent independent validation studies across general community cohorts have highlighted variable internal consistency estimates (Cronbach’s α typically ranging from 0.40 to 0.70) attributable to its deliberate multidimensional coverage and brief item length. Despite structural heterogeneity in factor analytic investigations, the AQ-10 remains one of the most widely implemented clinical triage tools globally for identifying adults who warrant comprehensive secondary-care neurodevelopmental assessment.

2. Keywords

Adult Autism Spectrum Quotient, AQ-10, autism screening, autism spectrum disorder, psychometrics, Autism Research Centre, primary care triage, neurodevelopmental assessment, clinical sensitivity, Baron-Cohen

3. Authors

The Adult Autism Spectrum Quotient (AQ-10) was developed by researchers based at the Autism Research Centre within the Department of Psychiatry at the University of Cambridge, United Kingdom:

• Carrie Allison, Ph.D. — Director of Strategy and Clinical Research Associate at the Autism Research Centre, Department of Psychiatry, University of Cambridge, Cambridge, United Kingdom.
• Bonnie Auyeung, Ph.D. — Professor of Psychobiology, School of Philosophy, Psychology and Language Sciences, University of Edinburgh, Edinburgh, United Kingdom; and Affiliated Researcher, Autism Research Centre, University of Cambridge.
• Sir Simon Baron-Cohen, Ph.D., FBA, FMedSci — Professor of Developmental Psychopathology, Director of the Autism Research Centre, Department of Psychiatry, University of Cambridge, and Fellow of Trinity College, Cambridge, United Kingdom.

Correspondence regarding the instrument’s scientific foundations is managed through the Autism Research Centre, Douglas House, 18b Trumpington Road, Cambridge CB2 8AH, United Kingdom.

4. Purpose

The primary purpose of the Adult Autism Spectrum Quotient (AQ-10) is to provide a rapid, user-friendly, and clinically accurate screening mechanism for identifying adults who may have undiagnosed Autism Spectrum Disorder (ASD) and who would benefit from referral for a comprehensive diagnostic evaluation. In routine clinical practice, primary care practitioners, general medical clinicians, and allied health professionals frequently encounter adults presenting with chronic interpersonal difficulties, social anxiety, depression, atypical occupational trajectories, or executive functioning challenges. However, full comprehensive assessments—utilizing gold-standard clinical instruments such as the Autism Diagnostic Observation Schedule (ADOS-2) and the Autism Diagnostic Interview-Revised (ADI-R)—are resource-intensive, requiring multiple hours of specialized administration and multidisciplinary evaluation.

Prior to the introduction of the AQ-10, clinicians often turned to the 50-item Autism Spectrum Quotient (AQ-50; Baron-Cohen et al., 2001). Although the AQ-50 is clinically validated, its administration time (approximately 10 to 15 minutes) is often prohibitive within standard 10-minute general practice consultations. In response to this operational bottleneck, the National Institute for Health and Care Excellence (NICE) commissioned the development of an abbreviated instrument capable of maintaining high discriminative validity while requiring fewer than three minutes to complete. The AQ-10 directly satisfies this requirement by reducing patient burden and streamlining the clinical decision-making pathway.

Beyond primary care referral triage, the AQ-10 serves extensive research applications. It is frequently employed in large-scale epidemiological investigations, population-level biobanks, and cognitive neuroscience paradigms where dimensional autism traits must be assessed rapidly alongside comprehensive phenotypic or genomic batteries. In academic research, it functions as a quantitative index of the broader autism phenotype (BAP), facilitating stratification without inducing participant fatigue. However, clinical guidelines and the instrument’s developers emphasize that the AQ-10 is purely a screening measure; it does not constitute an official diagnostic determination and cannot replace multidisciplinary clinical judgment.

5. Psychological Construct

The psychological construct evaluated by the AQ-10 is dimensional autistic traits in adult individuals of average or above-average intellectual functioning (intelligence quotient [IQ] ≥ 70). Autistic traits are conceptualized as continuous phenotypic dimensions distributed across both clinical and neurotypical populations. The AQ-10 samples 10 items representing five empirically delineated domains derived from the original AQ-50, allocating exactly two items per domain:

5.1 Attention Switching

Attention switching reflects cognitive flexibility, set-shifting ability, and executive control when managing concurrent tasks or sudden environmental changes. In neurotypical individuals, the redirection of attentional resources between competing sensory streams occurs relatively fluidly. Conversely, autistic cognitive profiles are characterized by high task absorption, monotropic focus, and distress or cognitive inertia during task transition. On the AQ-10, this construct is operationalized through items assessing multitasking capacity (Item 3: “I find it easy to do more than one thing at once”) and the speed of returning to an interrupted activity (Item 4: “If there is an interruption, I can switch back to what I was doing very quickly”).

5.2 Attention to Detail

Attention to detail pertains to perceptual processing style, specifically the propensity to focus on local features versus the holistic or global configuration of stimuli. Autistic individuals frequently exhibit enhanced sensory discrimination and a local processing bias. The AQ-10 assesses this tendency through self-reported auditory acuity (Item 1: “I often notice small sounds when others do not”) and preference for local components over global gestalts (Item 2: “I usually concentrate more on the whole picture, rather than the small details”, reverse-scored).

5.3 Communication

The communication dimension targets pragmatic language, the social use of verbal communication, and the comprehension of indirect linguistic cues. Pragmatic deficits in autism encompass challenges with conversational reciprocity, turn-taking, and decoding non-literal expressions such as irony, sarcasm, metaphor, and implicit subtext. In the AQ-10, pragmatic communication is captured through items addressing inferential verbal comprehension (Item 5: “I find it easy to ‘read between the lines’ when someone is talking to me”) and the monitoring of conversational partners’ engagement (Item 6: “I know how to tell if someone listening to me is getting bored”).

5.4 Imagination

Imagination in the context of the AQ framework refers specifically to social imagination and the capacity to infer fictitious internal mental states, narrative perspectives, and the thoughts and feelings of others within literary or conversational contexts. It is closely allied with Theory of Mind (ToM). The AQ-10 assesses this capacity through items measuring difficulty in deducing character motivations within literature (Item 7: “When I’m reading a story I find it difficult to work out the characters’ intentions”) and real-time inference of mental states from facial expressions (Item 9: “I find it easy to work out what someone is thinking or feeling just by looking at their face”).

5.5 Social Skill

The social skill domain evaluates natural interpersonal attunement, social interest, understanding of social conventions, and systemizing preferences within social interaction. The AQ-10 captures this construct through an item measuring strong categorization behaviors (Item 8: “I like to collect information about categories of things…”)—reflecting heightened systemizing tendencies over social engagement—and an item evaluating perceived difficulty in decoding interpersonal intentions during real-world social interaction (Item 10: “I find it difficult to work out people’s intentions”).

6. Theoretical Framework

The conceptual architecture of the AQ-10 is grounded in several foundational cognitive and neurodevelopmental theories formulated over the past three decades:

6.1 The Empathizing-Systemizing (E-S) Theory

Developed by Simon Baron-Cohen, the Empathizing-Systemizing (E-S) theory posits that psychological phenotypes can be understood through two continuous cognitive dimensions: Empathizing (E)—the drive to identify another person’s mental states, predict their behavior, and respond with an appropriate affective state; and Systemizing (S)—the drive to analyze systems, discover underlying rules governing systems, and construct predictive mechanical, taxonomic, or abstract rules. Autistic individuals typically present with a cognitive profile characterized by hyper-systemizing paired with hypo-empathizing (Type S or Extreme Type S profile). In the AQ-10, items tapping systemizing (Item 8) and attention to detail (Item 1) directly reflect this hyper-systemizing drive, whereas items indexing facial emotion reading (Item 9) and pragmatic decoding (Item 5) operationalize empathizing deficits.

6.2 Mindblindness and Theory of Mind (ToM) Deficits

The concept of Mindblindness (Baron-Cohen, 1995) postulates that autism involves an impairment in developing and deploying a Theory of Mind—the cognitive mechanism required to attribute mental states (beliefs, desires, intentions, knowledge) to oneself and others. Without a functioning mentalizing network, social interactions appear unpredictable, chaotic, and opaque. Items 5, 6, 7, 9, and 10 of the AQ-10 directly evaluate the everyday operationalization of ToM, querying the respondent’s conscious awareness of their capacity to parse intentionality, detect listener boredom, and infer unstated emotional states.

6.3 Weak Central Coherence (WCC) Theory

Formulated by Uta Frith and Francesca Happé, the Weak Central Coherence theory posits that typical neurocognitive processing is characterized by an innate drive to integrate disparate sensory information into higher-level, context-dependent meaning (central coherence). In contrast, autistic cognitive architecture exhibits a perceptual bias toward local, piecemeal detail at the expense of global integration. Item 2 of the AQ-10 (“I usually concentrate more on the whole picture, rather than the small details”) directly interrogates this perceptual processing divergence.

6.4 Monotropism and Executive Dysfunction

More recently, neurodiversity-affirmative and cognitive theories have framed attentional differences through the lens of monotropism (Murray, Lesser, & Lawson, 2005) and executive function models (Russell, 1997). Monotropic attention channels significant attentional energy into a singular processing stream, generating intense focus but creating substantial energetic costs during task-switching. Items 3 and 4 of the AQ-10 quantify the behavioral expression of this attentional allocation strategy.

7. Validity

The psychometric validity of the AQ-10 has been evaluated across clinical, community, and university populations with both supportive and cautionary findings:

7.1 Criterion and Diagnostic Validity

In the seminal validation study by Allison et al. (2012), the AQ-10 was derived using data from large cohorts: a clinical sample of adults diagnosed with ASD (n = 708) and a neurotypical comparison control group (n = 1,424). Items from the original AQ-50 were selected based on maximum discriminative capacity (high discriminative validity indices and high area under the receiver operating characteristic curve). Using a cutoff score of >6 (meaning a score of 7 or higher, although often operationalized in practice as ≥6 depending on local criteria):

• Sensitivity: 0.88 (0.85–0.91), indicating that 88% of confirmed autistic individuals met or exceeded the threshold.
• Specificity: 0.91 (0.89–0.93), indicating that 91% of neurotypical control participants correctly scored below threshold.
• Positive Predictive Value (PPV): 0.85, and Negative Predictive Value (NPV): 0.93 in the original study sample.
• Receiver Operating Characteristic (ROC): The Area Under the Curve (AUC) reached 0.92, reflecting outstanding discriminatory power between clinical and non-clinical populations.

7.2 External Replications in Secondary Mental Health Settings

Independent replication studies in real-world clinical referral settings have yielded more nuanced results. Booth et al. (2013) examined the performance of the AQ-10 among adult patients referred to specialized neurodevelopmental diagnostic services in the United Kingdom. In this enriched clinical sample—where all individuals were actively seeking diagnostic clarification—the AQ-10 maintained an AUC of approximately 0.86, with sensitivity remaining high (>0.80), but specificity declining to approximately 0.57 when discriminating ASD from other psychiatric conditions (such as severe affective disorders, personality disorders, and ADHD).

Similarly, Ashwood et al. (2016) evaluated the AQ-10 in a specialized diagnostic clinic (n = 250) and observed that while the instrument successfully flagged autistic individuals, high false-positive rates occurred among individuals with ADHD and complex trauma, who frequently endorsed items relating to attention switching and social cognitive strain. Consequently, clinicians are advised to consider the AQ-10 as a broad indicator of neurodevelopmental vulnerability rather than an autism-specific diagnostic instrument.

7.3 Convergent and Discriminant Validity

Convergent validity is evidenced by moderate-to-high correlations between the AQ-10 and other autism screening instruments. The AQ-10 correlates strongly with its parent measure, the AQ-50 (r = 0.80 to 0.86; Allison et al., 2012; Lundin et al., 2019), and demonstrates moderate positive associations with the Social Responsiveness Scale for Adults (SRS-2; r = 0.58–0.65). Discriminant validity against global measures of general intelligence is satisfactory (correlations with Full-Scale IQ typically falling between r = -0.05 and -0.12), demonstrating that AQ-10 scores are not confounded by general intellectual functioning in individuals without intellectual disability.

8. Reliability

The reliability profile of the AQ-10 illustrates classic psychometric trade-offs inherent in ultra-short scales:

8.1 Internal Consistency

Because the AQ-10 deliberately selects only two items from five distinct, heterogeneous cognitive domains, classical internal consistency coefficients based on the assumption of tau-equivalence (such as Cronbach’s alpha) tend to be modest. In the development sample of Allison et al. (2012), the overall internal consistency was reported as α = 0.72 in the combined clinical and control sample. However, subsequent independent investigations in non-clinical, general population samples have reported lower Cronbach’s alpha values, typically ranging from α = 0.40 to α = 0.61 (Booth et al., 2013; Lundin et al., 2019; Taylor et al., 2020).

When evaluated using McDonald’s omega total (ω_t), which does not assume equal factor loadings, reliability estimates improve moderately (ω_t ≈ 0.66–0.73). Psychometricians note that low internal consistency is an expected statistical consequence when an abbreviated scale captures broad, multifaceted constructs with minimal item redundancy. Rather than reflecting poor measurement quality, the lower alpha illustrates the breadth of the underlying autism phenotype sampled across just 10 questions.

8.2 Test-Retest Reliability

Test-retest stability of the AQ-10 has shown acceptable reproducibility over time. In non-clinical cohorts retested over intervals of 2 to 6 weeks, intra-class correlation coefficients (ICC) and Pearson correlation coefficients have ranged from r = 0.75 to 0.83 (Allison et al., 2012; Taylor et al., 2020), indicating that an individual’s self-reported trait endorsement remains stable across time in the absence of clinical intervention.

9. Factor Analysis

The structural dimensionality of the AQ-10 has generated considerable scholarly inquiry within quantitative psychometrics:

9.1 Exploratory Factor Analysis (EFA)

Initial exploratory analyses of the AQ-10 items suggested a multidimensional configuration loosely aligning with the five conceptual domains inherited from the AQ-50. However, because each domain contains only two items, standard EFA frequently encounters identification problems, Heywood cases, or unstable two-item factor solutions. When constrained to an exploratory bifactor framework, items generally display positive loadings on a broad, overarching general “Autistic Traits” factor, while specific residual variances cluster around social-communication versus attention/perceptual factors.

9.2 Confirmatory Factor Analysis (CFA)

Confirmatory investigations testing the original 5-factor correlated model have reported mixed fit metrics across different populations. In general adult community datasets (e.g., Lundin et al., 2019; N > 2,000):

• Unidimensional Model: A strict single-factor model typically yields sub-optimal fit indices: Root Mean Square Error of Approximation (RMSEA) ≈ 0.065–0.082; Comparative Fit Index (CFI) ≈ 0.78–0.88; Tucker-Lewis Index (TLI) ≈ 0.72–0.84. This demonstrates that autistic traits cannot be simplified into an entirely unidimensional construct without structural distortion.
• Two-Factor Model: Models specifying two correlated higher-order factors—(1) Social Communication and Interaction (Items 5, 6, 7, 9, 10) and (2) Focused Interests and Details / Attention Switching (Items 1, 2, 3, 4, 8)—consistently exhibit superior empirical fit: RMSEA ≈ 0.038–0.048; CFI ≈ 0.93–0.96; TLI ≈ 0.91–0.95. This two-factor bifurcation closely mirrors the modern dyadic diagnostic criteria established in the DSM-5 (Criterion A: Persistent deficits in social communication and social interaction; Criterion B: Restricted, repetitive patterns of behavior, interests, or activities).
• Bifactor Model: A bifactor configuration specifying one general factor alongside two group factors often provides the best statistical fit (CFI > 0.96, RMSEA < 0.04), supporting the practice of calculating a single composite total score for clinical screening purposes while acknowledging underlying multidimensionality.

10. Instrument / Measurement Tool

• Standard Designation: Adult Autism Spectrum Quotient (AQ-10) – Self-administered.
• Target Population: Adults (aged 16 years and older) with average or above-average cognitive ability (IQ ≥ 70).
• Administration Format: Self-report paper-and-pencil questionnaire, digital web-based form, or clinician-assisted interview.
• Administration Duration: Approximately 2 to 3 minutes.
• Total Item Count: 10 self-report statements.
• Response Scale: 4-point forced-choice response options:
  • Definitely Agree
  • Slightly Agree
  • Slightly Disagree
  • Definitely Disagree
• Scoring Architecture:
  • Items are scored using a collapsed binary system: each item yields either 0 or 1 point.
  • Direct Scoring (1 point for “Definitely Agree” or “Slightly Agree”): Items 1, 7, 8, and 10.
  • Reverse Scoring (1 point for “Definitely Disagree” or “Slightly Disagree”): Items 2, 3, 4, 5, 6, and 9.
  • Total Score Range: 0 to 10 points.
• Clinical Cutoff and Referral Threshold:
  • Score > 6 (i.e., 7 to 10 points, or ≥ 6 under sensitive screening guidelines): Significant endorsement of autistic traits; clinical guidelines (e.g., NICE CG142) recommend referral for a formal comprehensive diagnostic assessment by a specialist multidisciplinary team.
  • Score ≤ 6: Autistic traits are within the non-clinical range. However, clinical presentation must always take precedence over a negative screening score if functional impairment or clinical suspicion persists.

11. Permissions & Fee and Test Year

The Adult Autism Spectrum Quotient (AQ-10) was published in 2012 by Carrie Allison, Bonnie Auyeung, and Simon Baron-Cohen in the Journal of the American Academy of Child and Adolescent Psychiatry. The instrument was developed at the Autism Research Centre (ARC), University of Cambridge.

The AQ-10 is designated as an open-access, non-commercial clinical and research assessment tool. In accordance with the academic dissemination policies of the Autism Research Centre, the instrument is available free of charge for non-commercial research, educational, and clinical screening purposes. Clinicians and researchers may administer and reproduce the scale without paying licensing fees, provided that appropriate bibliographic citation is accorded to the developers and the source publication. Any commercial re-distribution, automated software integration into commercial clinical systems, or proprietary resale requires formal written permission from the Autism Research Centre, Cambridge.

12. References

• Allison, C., Auyeung, B., & Baron-Cohen, S. (2012). Toward brief “red flags” for autism spectrum conditions: The Short Autism Spectrum Quotient and the Short Quantitative Checklist in 1,000 cases and 3,000 controls. Journal of the American Academy of Child & Adolescent Psychiatry, 51(2), 202–212. https://doi.org/10.1016/j.jaac.2011.11.003
• Ashwood, K. L., Gillan, N., Bramham, J., Liang, H., Sadik, S., Craig, M. C., & Murphy, D. G. (2016). Predicting the diagnosis of autism in adults using the Autism-Spectrum Quotient (AQ) sensitivity and specificity in clinical practice. Autism Research, 9(2), 244–253. https://doi.org/10.1002/aur.1515
• Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory of mind. MIT Press.
• Baron-Cohen, S., Wheelwright, S., Skinner, R., Martin, J., & Clubley, E. (2001). The Autism-Spectrum Quotient (AQ): Evidence from Asperger syndrome/high-functioning autism, males and females, scientists and mathematicians. Journal of Autism and Developmental Disorders, 31(1), 5–17. https://doi.org/10.1023/A:1005653411471
• Booth, T., Murray, A. L., McKenzie, K., Kuenssberg, R., O’Donnell, M., & Burnett, H. (2013). Brief report: An evaluation of the AQ-10 as a brief screening tool for autism spectrum disorders. Journal of Autism and Developmental Disorders, 43(12), 2997–3000. https://doi.org/10.1007/s10803-013-1844-5
• Frith, U. (1989). Autism: Explaining the enigma. Basil Blackwell.
• Lundin, A., Kosidou, K., & Dalman, C. (2019). Measuring autism traits in the general population: A psychometric study of the Autism-Spectrum Quotient 10-item version (AQ-10) in a large Swedish population sample. Psychiatry Research, 274, 303–308. https://doi.org/10.1016/j.psychres.2019.02.046
• Murray, D., Lesser, M., & Lawson, W. (2005). Attention, monotropism and the diagnostic criteria for autism. Autism, 9(2), 139–156. https://doi.org/10.1177/1362361305051398
• National Institute for Health and Care Excellence (NICE). (2012). Autism spectrum disorder in adults: Diagnosis and management (NICE Clinical Guideline CG142). London: National Institute for Health and Care Excellence. https://www.nice.org.uk/guidance/cg142
• Taylor, E. C., Livingston, L. A., Clutterbuck, R. A., & Shah, P. (2020). Psychometric evaluation of the Autism-Spectrum Quotient (AQ-10) in autistic and non-autistic adults. Molecular Autism, 11, Article 97. https://doi.org/10.1186/s13229-020-00398-3

13. Items of the Scale (Questionnaire)