Module 4 — GQ-ASC — The Autism Questionnaire Written for Women

 

Why the standard description of autism was drawn from boys, what the GQ-ASC asks instead, where it came from, and an honest account of how much weight a score of 57 can carry.

 

Autistic Self-Discovery · Part Two — The Broad Screeners · 14 min read, about 26 min with the workbook

 

The big idea

 

If you have spent years collecting other explanations — anxious, shy, sensitive, difficult, gifted but lazy, too much, not enough — this module is about a questionnaire that was written because the standard description of autism was drawn from somebody else, and about how much weight that questionnaire can actually bear.

 

One study, by the people who made the instrument, with no clinician confirming anybody's diagnosis. That is not a scandal — it is how most instruments start. It is a problem only when the instrument is then talked about as though the checking had already happened.

 

Step 1 — The lesson

 

A. The picture in the book

 

For most of the time autism has had a name, the description of it was assembled from the children who were being sent for assessment. Those children were overwhelmingly boys. The traits that made it into the textbooks — and from there into the assessment tools — were the traits of the people already in the room.

 

Clinicians running diagnostic services started saying so out loud around 2011: that girls and women were being missed not because they were less autistic but because the questions were framed around a presentation that was not theirs, and that reworded questions surfaced them.¹

 

Diagram — A · The picture in the book. The description was accurate about the children it was drawn from. That is a different thing from being complete. Everyone who did not resemble the drawing had to wait for somebody to redraw it.

 

What happened to the people who were missed is documented. Women diagnosed in adulthood describe years of pretending to be normal, of professionals reading their gender before their difficulties, and of a conflict between being autistic and the version of femininity they were expected to perform.² A 2026 review pooling fourteen studies found a systematic tilt toward detecting the more externalising, more stereotypically male profile — and found that specialist assessment did not correct it.³ That review comes with a heavy caveat we will return to, but the direction of the finding is not seriously disputed.

 

So there was a real gap, and there was a real reason to want an instrument that asked different questions. The GQ-ASC is one attempt at that. This module is about how good an attempt it is.

 

B. Five areas, one number

 

The version you will take asks a short set of statements and gives you a single total. There are no subscale scores to read — unlike the CAT-Q in Module 5, everything collapses into one number.

 

The items were grouped by the researchers into five areas.⁴ Knowing what they are is useful even though you will not get five separate numbers back, because it tells you what the questionnaire is looking at — and, just as importantly, what it is not.

 

Diagram — B · Five areas, one number. Imagination and play, camouflaging, sensory sensitivities, socialising, interests. Five areas feeding one total. Together they explained about forty per cent of the variation between people in the study, which means the majority of what makes one respondent different from another is not captured here.

 

Imagination and play asks about the interior life — invented worlds, elaborate pretend, escaping into fiction. This is the area most obviously written about a child.

 

Camouflaging asks whether you study other people in order to copy them. This is the area with the most independent support behind it, and Module 5 is entirely about it.

 

Sensory sensitivities asks about noise, texture, light, smell. Socialising asks about friendship, groups, and finding one person easier than five. Interests asks about intensity and duration rather than about trains.

 

You may already notice something. Those are recognizable autistic experiences, but three of the five are not gendered claims at all — sensory difference, socialising and intense interests are asked about by every autistic-traits questionnaire ever written. What is distinctive here is the framing of the questions, not a separate list of female traits.

 

C. Shortened, not re-cut

 

Where an instrument came from matters, because it explains the shape it is still in.

 

The original was the Q-ASC, put together in 2011 by Tony Attwood, Michelle Garnett and Agnieszka Rynkiewicz. It was a questionnaire about girls, aged roughly five to nineteen, filled in by a parent. It was never published as a study — it exists as a measurement instrument with no sample, no psychometrics and no peer review of its own.⁵

 

Diagram — C · Shortened, not re-cut. The adult questionnaire is the children's questionnaire, shortened and re-worded so that you answer it about yourself instead of a parent answering it about you. That is a reasonable way to build a first draft. It is not the same as designing an instrument for adults from the beginning.

 

In 2018 it got its first real study. Parents of 238 autistic children and adolescents filled it in, and the analysis found interpretable structure and differences between groups.⁶ But there was no comparison group of non-autistic children — everyone in the study was already diagnosed. A study with no comparison group can describe the people in it. It cannot tell you whether the questionnaire distinguishes them from anybody else, and so it produced no cut-off score, no sensitivity and no specificity. None were possible.

 

The adult self-report form arrived in 2020: twenty-one of the items kept, re-worded to the first person, answered by 672 women aged 18 to 72.⁴ That study is the entire adult evidence base. There is not a second one.

 

D. What the number 57 is actually telling you

 

The screener bands your total, and the high band begins at 57. That number is not arbitrary: it comes from the 2020 study, where a line drawn just above 56 sorted the autistic and non-autistic women in the sample better than any other line.⁴

 

Here is what that line achieved. It put 80 per cent of the 672 women on the correct side. Which means it put one in five on the wrong side.

 

Diagram — D · What the number 57 is actually telling you. Eighty per cent correct is a real result and it is also a coin that lands wrong twenty times in a hundred. Before you read your own score as an answer, sit with the fact that twenty of these women got a number that pointed away from their own reality.

 

Two things about that 80 per cent are worth knowing, and you will not find either on most pages about this questionnaire.

 

It is not sensitivity. The scoring sheet that circulates with the adult form describes the cut-off as "sensitive to 80% of cases." That is a misdescription. Eighty per cent is the overall proportion of everybody sorted correctly, autistic and non-autistic pooled together. A separate sensitivity figure — how many autistic women it catches — has never been published, and neither has specificity, and neither has an area-under-the-curve value.

 

The errors do not appear to be evenly split. The one independent evaluation that exists compared the AQ-50, the GQ-ASC and the CAT-Q in 343 autistic and 345 non-autistic women. It found the GQ-ASC identified autistic women about as well as the CAT-Q did, but that its specificity was poor: a lot of non-autistic women also scored above the line.⁷ That study is a preprint and has not been through peer review, so treat it as a signal rather than a settled finding. It is, however, the only independent look anyone has taken.

 

Put plainly: a high score here is a reason to keep going. It is not evidence of much on its own, and a low score is not a reason to stop — particularly if you recognized yourself in Module 5.

 

INSIDE THE INSTRUMENT

 

GQ-ASC — Girls Questionnaire for Autism Spectrum Condition, adult self-report form

 

Where it came from

 

Attwood, Garnett and Rynkiewicz (2011), Questionnaire for Autism Spectrum Conditions (Q-ASC) — an unpublished parent-report measurement instrument about girls, carrying no sample, no psychometric data and no peer review of its own.⁵ The circulating child form covers ages 5 to 12 with 54 items and no scoring key. Every psychometric claim made about the Q-ASC therefore rests on later studies rather than on its development.

 

What it is made of

 

The adult self-report form scores 21 items on a four-point Likert scale (definitely disagree to definitely agree), six of them reverse-scored, giving a published range of 21 to 84. Principal components analysis produced five components — imagination and play, camouflaging, sensory sensitivities, socializing, interests — accounting for 40.40% of total variance. There are no subscale scores in use; only the total is scored.⁴

 

How well it performs

 

The 2018 parent-report study analyzed 238 children and adolescents with a clinical diagnosis of ASD Level 1 without intellectual or language impairment, producing eight interpretable components. It had no non-autistic comparison group, so no sensitivity, specificity, AUC or cut-off was or could be derived.⁶ Note for anyone searching: this paper is universally cited as Ormond et al. but is now indexed on PubMed and SpringerLink under first author Simcoe, following a post-publication name change.

 

The adult validation recruited 672 cisgender and trans women aged 18 to 72 online — 350 autistic, 322 non-autistic. A cut-off of 57 correctly classified 80.0% of cases. ROC performance is described only as an "excellent level of discrimination"; no AUC value, no sensitivity and no specificity is published.⁴

 

No test-retest reliability has ever been published for the adult form. No norms exist — no normative sample, no percentile tables, no demographic stratification.

 

Where it was validated

 

Once, by its own developers, in a self-selected online sample. Group membership rested on the participant's own reported diagnosis plus an Autism Quotient screen; no clinician-administered diagnostic assessment appears anywhere in the design, and there was no cross-validation sample. The 80.0% is therefore an in-sample classification rate, not an out-of-sample accuracy estimate.⁴

 

What it cannot do

 

It has no independent peer-reviewed validation. The only independent evaluation is an unpublished preprint (343 autistic and 345 non-autistic women) which found detection comparable to the CAT-Q but poor specificity, with many non-autistic women scoring above the cut-off.⁷ Its exact sensitivity, specificity and AUC values are not retrievable from any accessible copy.

 

Its measured content validity is low. A 2026 two-round modified Delphi with 66 raters (33 autistic women, 33 academics and clinicians) found only 8 of the 21 items (38.1%) reached excellent agreement on relevance. Scale-level content validity was 0.66 among autistic women and 0.71 among professionals, neither meeting the threshold for excellent; the GQ-ASC tied with the AQ for the lowest relevance of the five measures tested, and the imagination and play items were flagged for revision or removal.⁸

 

The sex-difference premise is only partly supported. A 2024 meta-analysis of narrow restricted and repetitive behavior constructs found autistic males higher on stereotyped behaviors (SMD 0.21) and restricted interests (SMD 0.18), but no significant sex difference in sensory experiences (SMD −0.09) or insistence on sameness (SMD 0.01).⁹ Two of the GQ-ASC's five components sit in domains where the sex difference is either absent or very small.

 

The construct it is built on is contested. A discourse analysis of 83 clinical texts names the circularity directly: a woman who is not stereotypically feminine may be read as having an autistic profile, and a woman who resembles her peers may be read as masking, so every observation confirms the hypothesis.¹⁰ An instrument built on an unfalsifiable construct inherits that problem.

 

Caution on the gender-bias meta-analysis. The 2026 pooled estimate of 0.29 (95% CI 0.11 to 0.47) is directional at best: heterogeneity was essentially total (I² = 99.93%), the prediction interval ran from −0.40 to 0.97, and funnel-plot asymmetry was significant.³ Cite the direction, not the magnitude.

 

What an adequately reported instrument looks like

 

For contrast, the Autistic Women's Experience scale — the same category of tool, developed for the same purpose — reports Cronbach's alpha 0.946, test-retest ICC 0.764, ROC-AUC 0.925 overall and 0.936 in women, and a cut-off giving 88% sensitivity and 85% specificity in 153 autistic and 489 general-population adults.¹¹ Those are in-sample figures from its own development team and await replication too — but they are reported. The GQ-ASC's equivalents do not exist.

 

What a clinician does with it

 

Use it as a structured interview prompt, not as a screen with an operating characteristic. Its value is in the questions it asks that male-normed instruments do not, and in what a client says when asked them. Clinical reviews aimed at practitioners describe the GQ-ASC as showing "better performances" with women;¹² note when reading them that such reviews are sometimes co-authored by the instrument's developers and that no comparative accuracy data supports the claim.

 

Validity tier: 2 — published, but what it measures and how well are both contested. It is not a Tier 1 instrument. It has one non-independent validation, no norms, no test-retest, no published sensitivity or specificity, low measured content validity, and a single independent preprint suggesting weak specificity.

 

Diagram — E · The spec sheet has blanks. This is not a verdict on the questionnaire. It is a list of the things nobody has measured yet. An instrument with this many blanks can still be a good conversation. It cannot be a decision.

 

STRENGTHS LENS

 

It asks about things nobody else was asking about.

 

When 343 autistic and 345 non-autistic women were given the AQ-50, the GQ-ASC and the CAT-Q side by side, the autistic participants rated the AQ-50 as the most stereotyped and least interpretable of the three, and preferred the GQ-ASC.⁷ That is worth something the psychometrics do not capture: an instrument you can answer honestly gets better data than one you spend the whole time translating.

 

And the questions themselves do real work. Asking a forty-year-old woman about invented worlds, about studying people in order to copy them, about which of five people in a room she can actually hear — these open doors that "do you have difficulty with social interaction" leaves shut. Take the questions seriously even while you hold the number loosely.

 

F. What helps

 

The most useful thing this questionnaire produces is not the total. It is the specific items you read twice.

 

1. Mark the items that landed, not just the score.

 

Go back through and note which statements made you stop. Three items you strongly recognized carry more information for a conversation with a clinician than a total does, and the total is the part that has never been independently validated. The workbook below is built to capture exactly this.

 

2. If your score is low and you still recognize yourself, do not discard yourself.

 

This instrument was written about girls and adapted for adults, and autistic women who were diagnosed later in life have specifically said its framing does not fit them. A low score in someone who has spent thirty years getting good at not being noticeable is a plausible false negative, not an answer.

 

3. Take a second screener that measures something different.

 

Because a single instrument with unpublished specificity is thin ground. The CAT-Q in Module 5 measures the camouflaging that may be suppressing your score here. The RAADS-14 in Module 2 is short and asks a different set of questions. Agreement across instruments that share no items is far more informative than one high number.

 

4. Write the history down before you go anywhere with it.

 

Women diagnosed after forty describe years of being handed other labels first — anxiety, depression, an eating disorder, a personality disorder.¹³ The pattern over time is often more diagnostic than any present-day questionnaire. School reports, old letters, what your family says you were like at six.

 

5. Bring the disagreement with you.

 

If your score contradicts your own sense of yourself, that contradiction is information for an assessing clinician, not an embarrassment to hide. Say it directly: the questionnaire put me here, and this is why I do not think that is right.

 

Step 2 — Take the screener

 

Free and confidential, about five to seven minutes. You will get a single total between 0 and 84 — no subscales. Read section D first if you can: this questionnaire has one validation study behind it and no published sensitivity, specificity or norms, so the items you recognize are worth more than the number they add up to.

 

Take the GQ-ASC

 

Before you start

 

The GQ-ASC began as a parent-report questionnaire about girls; the adult self-report form has been validated once, by its own developers, with no clinician confirming any participant's diagnosis. It is a screen and a conversation starter, not a diagnosis — only a clinician can diagnose. No score here means you are or are not autistic, and the one independent evaluation to date found that many non-autistic women also score above the cut-off.

 

Step 3 — Your workbook

 

Your answers save to this device only — we cannot see a word of what you write. This module lets you record your total if you took the screener, then do the part that matters more: capture the specific items you recognized, whether the number matches your own sense of yourself, and the history you would want an assessing clinician to hear first.

 

Your GQ-ASC result

 

Took the GQ-ASC? Put your total in below to see where it lands. Entirely optional — skip it if you would rather just read.

 

Score bands: 0–28 = Low or no autistic traits; 29–56 = Moderate autistic traits; 57–84 = High autistic traits

 

Fields: GQ-ASC · Girls Questionnaire for Autism Spectrum Condition; Total score (0–84); My total (enter 0–84)

 

The items that landed

 

Section F, item 1. The specific statements you stopped on carry more information than the total does — and unlike the total, they have not been through anybody's statistics.

 

Fields: Which statements made me stop and read again; What each one actually looks like in my life

 

Does the number match you?

 

If it does not, that disagreement is information — not something to hide.

 

Fields: My score against my own sense of myself: It fits / It is lower than I expected / It is higher than I expected / I do not know what I expected; Why I think that is

 

The five areas, one at a time

 

You only get one number back, so this is where you unpack it. Which of the five is loudest for you?

 

Fields: The area that describes me most: Imagination and play / Camouflaging / Sensory sensitivities / Socialising / Interests; An example of it, from this month

 

The history, before you go anywhere with it

 

Section F, item 4. The pattern over time is often worth more to an assessing clinician than any present-day score.

 

Fields: Labels I was given before anyone said the word autistic; What people who knew me at six or seven say I was like

 

What I would take to an assessment

 

Not the score. Three things you would want somebody to know first.

 

Fields: The three things

 

Appendix — Research companion

 

Peer-reviewed research

 

2. Bargiela S, Steward R, Mandy W (2016). The experiences of late-diagnosed women with autism spectrum conditions: an investigation of the female autism phenotype. Journal of Autism and Developmental Disorders, 46(10), 3281-3294. DOI 10.1007/s10803-016-2872-8. View the paper Framework analysis of in-depth interviews with 14 women aged 22 to 30 diagnosed in late adolescence or adulthood. Participants described pretending to be normal, gender leading professionals to miss their autism, and conflict between autism and a traditional feminine identity; experiences of sexual abuse were widespread in the sample. This is the most-cited empirical anchor for the camouflaging and delayed-diagnosis account. Limitation: 14 self-selected, verbally fluent, late-diagnosed women in a narrow age band cannot establish that a distinct female phenotype exists at population level, only that these women's diagnostic pathways were delayed.

 

4. Brown CM, Attwood T, Garnett M, Stokes MA (2020). Am I autistic? Utility of the Girls Questionnaire for Autism Spectrum Condition as an autism assessment in adult women. Autism in Adulthood, 2(3), 216-226. DOI 10.1089/aut.2019.0054. View the paper The only validation of an adult self-report form of the GQ-ASC. 672 cisgender and trans women aged 18 to 72 were recruited online, 350 autistic and 322 non-autistic. Principal components analysis gave 5 components (imagination and play, camouflaging, sensory sensitivities, socializing, interests) accounting for 40.40% of total variance, all differing significantly between groups, and a cut-off of 57 correctly identified 80.0% of cases; no AUC value, sensitivity or specificity is reported. Group membership rested on participants' own reported diagnosis plus an Autism Quotient screen, with no clinician-administered diagnostic assessment anywhere in the design. Limitation: with no clinical gold standard, a self-selected online sample and no cross-validation sample, the 80.0% figure is an in-sample classification rate rather than an out-of-sample accuracy estimate.

 

9. Edwards H, Wright S, Sargeant C, Cortese S, Wood-Downie H (2024). Research review: a systematic review and meta-analysis of sex differences in narrow constructs of restricted and repetitive behaviours and interests in autistic children, adolescents, and adults. Journal of Child Psychology and Psychiatry, 65(1), 4-17. DOI 10.1111/jcpp.13855. View the paper Forty-six studies were narratively synthesised and 25 entered four random-effects meta-analyses of sex differences in narrowly defined restricted and repetitive behaviours in autistic people. Autistic males scored higher on stereotyped behaviours (SMD 0.21, 95% CI 0.09 to 0.33) and restricted interests (SMD 0.18, 95% CI 0.07 to 0.29), but there was no significant sex difference in sensory experiences (SMD -0.09) or insistence on sameness (SMD 0.01). This directly qualifies an instrument that devotes whole components to sensory sensitivities and to interests. Limitation: the significant effects are small, and most included studies used male-normed measures, so measurement bias could be generating or concealing the very differences being estimated.

 

11. Groen Y, Ebert WM, Dittner FM, Stapert AF, Henning D, Greaves-Lord K, Davids RCDL, Castelein S, Baron-Cohen S, Allison C, Van Balkom IDC, Piening S (2023). Measuring the Autistic Women's Experience (AWE). International Journal of Environmental Research and Public Health, 20(24), 7148. DOI 10.3390/ijerph20247148. View the paper A directly comparable female-focused adult self-report instrument, included here as a benchmark for what the GQ-ASC does not report. 153 autistic adults (85 women, 68 men) and 489 general-population adults completed the 49-item AWE, which reports Cronbach's alpha 0.946, test-retest ICC 0.764 in a 70-person retest subsample, ROC-AUC 0.925 overall and 0.936 in women, and an optimal cut-off of 123 on a 49 to 196 range giving 88% sensitivity and 85% specificity. Limitation: these are in-sample figures from the instrument's own development team in a single Dutch-language sample, so they await independent replication exactly as the GQ-ASC's do.

 

10. Moore I, Morgan G, Howard C (2025). Constructions of female autism in professional practices: a Foucauldian discourse analysis. Feminism & Psychology, 35(2), 206-227. DOI 10.1177/09593535241283325. View the paper A Foucauldian discourse analysis of 83 texts (clinical reports, training slides, leaflets, articles and books) produced by 10 UK-based clinicians, examining how female autism is constructed professionally. It finds the construct framed as a medical-scientific advance and a gender-equality gain while in practice extending the expert gaze and reinforcing binary gender norms, and it names the circularity directly: a woman who is not stereotypically feminine may be read as having an autistic profile, while one who resembles her peers may be read as masking. Limitation: a critical discourse analysis of UK professional texts makes no empirical claim about whether female-presentation instruments detect autism, so it undercuts the construct's coherence rather than the GQ-ASC's accuracy.

 

6. Simcoe SM, Brownlow C, Garnett MS, Rynkiewicz A, Attwood T (2018). Profiling autism symptomatology: an exploration of the Q-ASC parental report scale in capturing sex differences in autism. Journal of Autism and Developmental Disorders, 48(2), 389-403. DOI 10.1007/s10803-017-3324-9. View the paper Parents of 238 children and adolescents with a clinical diagnosis of ASD Level 1 without intellectual or language impairment completed the Q-ASC; principal components analysis produced 8 interpretable components with significant mean differences by age and gender group. There was no non-autistic comparison group, so no sensitivity, specificity, AUC or cut-off was or could be calculated. This paper is universally cited as Ormond et al. 2018 but is now indexed on PubMed and SpringerLink under first author Simcoe following a post-publication author name change; the DOI and page range are unchanged. Limitation: a single-group clinical sample with no controls cannot establish discriminant or diagnostic validity, so the eight components are descriptive structure only.

 

8. Uglik-Marucha N, Fredman J, Al-Shurafa T, Ragrag C, Happe F, Belcher H, Vitoratou S (2026). Are items from current self-report measures adequate for assessing autistic traits in women? Insights from a modified Delphi study with autistic women and professional experts. Molecular Autism, 2026; volume and article number not yet assigned at time of verification. DOI 10.1186/s13229-026-00724-1. View the paper A two-round modified Delphi with 66 participants (33 autistic women and 33 academics and clinicians) rated item relevance across the AQ-50, BAPQ, CATI, GQ-ASC and RAADS-14. Across the 21 GQ-ASC items only 8 (38.1%) achieved excellent agreement on relevance; scale-level content validity was 0.66 among autistic women and 0.71 among professionals, neither meeting the threshold for excellent content validity, and the GQ-ASC tied with the AQ for lowest relevance of the five measures. Imagination and play items were flagged for revision or removal. Limitation: this establishes content validity only, so it says nothing about diagnostic accuracy, and a 66-person expert panel recruited through specialist networks is not a representative sample of autistic women.

 

Clinical frameworks and position statements

 

5. Attwood T, Garnett MS, Rynkiewicz A (2011). Questionnaire for Autism Spectrum Conditions (Q-ASC). Unpublished measurement instrument; no journal, volume, issue or pages exist for this item. View the source The origin of the instrument is a parent-report screening questionnaire about girls, not a published study; it is cited in the literature only as a measurement instrument. No sample size, no psychometric data and no peer review attach to the 2011 item. The circulating child form covers girls aged 5 to 12 with 54 items and carries no scoring key or cut-off. Limitation: an unpublished instrument with no empirical evidence of any kind behind it, so every psychometric claim about the Q-ASC rests entirely on later studies rather than on its own development.

 

1. Gould J, Ashton-Smith J (2011). Missed diagnosis or misdiagnosis? Girls and women on the autism spectrum. Good Autism Practice, 12(1), 34-41. View the source The foundational clinical practice paper, written by clinicians at the UK National Autistic Society diagnostic service, arguing that girls and women are systematically missed because diagnostic questions are framed around male presentations, and proposing reworded probe questions to surface female presentations. Its influence is the reason female-presentation instruments such as the Q-ASC were built at all. Limitation: expert clinical opinion with no sample, no data and no psychometric content, so it can support the rationale for a female-sensitive tool but cannot support any claim that such tools work.

 

12. Rynkiewicz A, Zheng S, Lacroix A (2024). Special considerations for assessing and caring for autism in girls and women. Current Opinion in Psychiatry, 37(2), 71-77. DOI 10.1097/YCO.0000000000000913. View the source A narrative clinical review covering measurement bias in diagnostic tools, physical and mental health across puberty, motherhood and menopause, and employment. It states that conventional diagnostic tools struggle to capture autism in females while more sensitive instruments such as the GQ-ASC show better performances, and it is the clearest example of how the instrument is presented to practising clinicians. Limitation: a non-systematic narrative review co-authored by a co-developer of the Q-ASC, in which the claim of better performance is asserted without any comparative accuracy data, so it should be read as advocacy rather than evidence.

 

Lived experience

 

13. Leedham A, Thompson AR, Smith R, Freeth M (2020). I was exhausted trying to figure it out: the experiences of females receiving an autism diagnosis in middle to late adulthood. Autism, 24(1), 135-146. DOI 10.1177/1362361319853442. View the source Semi-structured interviews with 11 autistic women all diagnosed after the age of 40, analysed with Interpretative Phenomenological Analysis, yielding four superordinate themes: a hidden condition, the process of acceptance, the impact of others post-diagnosis, and a new identity on the autism spectrum. Participants described years of mislabelling and mental health misdiagnosis before recognition, and diagnosis as a shift from self-criticism to self-compassion that was nonetheless painful to absorb so late. Limitation: 11 UK participants, all diagnosed over 40 and self-selected through autism networks, so the account is vivid but not generalisable and carries recall bias.

 

Emerging or contested

 

3. Muino Tato A, Marques-Donoso A, Sanchez-Huete JC (2026). Gender bias in ASD diagnostic and screening tools: a systematic review and meta-analysis. Psychiatry International, 7(4), 145. DOI 10.3390/psychiatryint7040145. View the paper A PRISMA 2020 review of studies from 2015 to 2025 pooling 14 empirical studies with sample sizes from 55 to approximately 3,000,000, using Fisher r-to-z effect sizes in a random-effects model. The pooled effect was 0.29 (95% CI 0.11 to 0.47, p = 0.001), interpreted as systematic bias favouring detection of externalising, prototypically male profiles across both screening and formal evaluation, and not corrected by specialist assessment. The GQ-ASC is not among the instruments reviewed. Limitation: heterogeneity was essentially total (I2 = 99.93%) with a prediction interval of -0.40 to 0.97 and significant funnel-plot asymmetry, so the pooled estimate is directional at best and its magnitude should not be quoted.

 

7. Parsons CH, Jackson MC (2025). These questions are geared towards male autistics, not female: assessing the efficacy of autism questionnaires in identifying adult autistic women. PsyArXiv preprint, version 2 posted 26 March 2025; not peer reviewed, no journal, volume or pages. DOI 10.31234/osf.io/k4ey5_v2. View the paper The only independent evaluation of the adult GQ-ASC: 343 autistic and 345 non-autistic women completed the AQ-50, the GQ-ASC and the CAT-Q. The AQ-50 showed lower sensitivity but strong specificity in women, the CAT-Q and GQ-ASC identified autistic women about equally well, and critically the GQ-ASC showed low specificity, with many non-autistic women scoring above its cut-off. Autistic participants rated the AQ-50 as the most stereotyped and least interpretable and preferred the GQ-ASC, while still raising concerns that it is not age-appropriate for adults. Limitation: an unpublished, non-peer-reviewed preprint whose exact sensitivity, specificity and AUC values are not retrievable from any accessible copy.

 

Further reading — general background

 

Uglik-Marucha N, Show S, Vitoratou S, Happe F, Belcher H (2026). I fit the category of the box, it just doesn't describe me well: exploring the perspectives of autistic women and gender-diverse individuals on self-report autism measures. PLOS ONE, 21(1), e0337600. DOI 10.1371/journal.pone.0337600. View the source Interviews, focus groups and surveys with 22 autistic women and AFAB gender-diverse individuals, analysed with reflexive thematic analysis, producing two overarching themes: that questionnaires measure only one way to be autistic and not in an autism-friendly manner, and that key lived experiences are missing. The authors conclude the measures have non-satisfactory content validity for this group. Note for accurate reading: this study examined the AQ-10, RAADS-14 and BAPQ only, and its remark about the GQ-ASC being inappropriate for adults appears in the introduction as a summary of a prior study rather than as its own participant data. Limitation: 22 participants, none of whom were asked about the GQ-ASC.

 

Peer-reviewed = checked by independent experts before publication. Clinical model = an established professional framework, not a single study.

 

Up next

 

Module 5 - CAT-Q — Measuring Autistic Masking and Camouflaging

 

All modules in Autistic Self-Discovery

 

A questionnaire cannot know you. A person can. This course was built by clinicians who are part of the New Path Family. If you are holding a score that does not match your life, or a lifetime of labels that never quite fit, that is worth talking through with someone who has heard it before. Therapy for clients in California and coaching worldwide, all by telehealth, are offered by our sister company New Path Family of Therapy Centers, Inc. A conversation costs nothing and there is no pressure. Saving this for later counts too. Talk with the New Path team