The testing paradox — girls pass direct assessment, but fail parent questionnaires
Research from the University of Huddersfield examined 79 children with autism aged 6-12 — 20 girls, 59 boys — and found something diagnostically significant.
Girls scored lower in restricted, repetitive and stereotyped behaviours than boys. Specifically, they demonstrated fewer stereotyped behaviours and restricted interests measurable through ADOS-2 assessment.
But on parent-rated questionnaires measuring executive function — the cognitive processes governing goal-directed behaviour, attention regulation, and emotional control — girls showed significantly worse deficits than boys.
The divergence is sharp. Girls scored more severely on emotion regulation, cognitive regulation, and specific clinical scales measuring cognitive flexibility (Shift) and task initiation (Initiate).
Yet when researchers tested executive function directly using the Wisconsin Card Sorting Test — a standard measure of “cognitive flexibility” — they found no differences between girls and boys. Both groups showed comparable deficits in errors, perseverative errors, and reaction times.
The pattern: girls appear more impaired when parents assess daily functioning, but perform equivalently to boys on direct clinical testing.
The researchers suggest this divergence may reflect parental bias — parents expect greater cognitive and emotional regulation from daughters than sons, leading to more critical parental assessments of autistic girls.
Translation (if needed): the same struggle receives harsher evaluation when displayed by a daughter.
The executive function gaps that testing cannot capture
The Wisconsin Card Sorting Test measures cognitive processes in controlled clinical conditions. The child sits with an assessor. The environment is structured. The task has clear parameters. Performance occurs in time-limited sessions.
Parent questionnaires like the BRIEF-2 measure executive function as it operates across contexts throughout daily life: getting ready for school; managing homework; regulating emotional responses during unstructured social situations; initiating tasks without external prompting; and shifting between activities when routines change.
These are precisely the domains where compensation and masking operate most invisibly to external observers.
An autistic girl may successfully complete the Wisconsin Card Sorting Test through intense focused effort in a controlled setting. The assessor sees adequate performance and infers adequate underlying cognitive processes.
But maintaining that performance level across an entire school day, through multiple transitions, whilst processing social expectations and managing sensory processing input, requires sustained effort that leaves no cognitive bandwidth for the flexibility and emotional regulation parents observe deteriorating at home.
The compensation is successful in the clinic. It collapses under sustained real-world demands.
What parents rate as worse “executive function” may actually be evidence of greater compensatory effort producing functional breakdown that testing doesn’t capture because testing samples performance, not sustained capacity.
The study found no differences in intelligence between girls and boys. Both groups scored in the 72-132 IQ range with comparable means. The girls demonstrating worse executive function deficits on parent ratings weren’t less capable. They were managing equivalent cognitive demands with equivalent cognitive resources whilst also managing the additional load of compensation strategies that boys either didn’t employ or employed less extensively.
The masking paradox from our article on shadow epistemology in neurodiversity applies here. Successful compensation renders neurodivergent phenomenology invisible to clinical observation precisely because it’s most successful. The cost of that compensation — the executive function and emotional regulation collapse parents observe — only becomes visible in contexts where clinical assessors are not present.
When gendered interests create diagnostic invisibility
The study found girls scored significantly lower in restricted, repetitive and stereotyped patterns of behaviour. But researchers noted something qualitative beyond the quantitative scores.
During clinical assessment, the researchers reported that several girls exhibited special interests that are more traditionally girl-like: arts and crafts, hairstyles and fashion, and specific animals (like foxes).
These interests met criteria for autistic intensity and pervasiveness. They dominated the girls’ attention and conversation. They served the same regulatory and absorptive functions that stereotypically “autistic” interests like trains or dinosaurs serve for boys (sorry boys, and I’m not doing us any favours by finding the latest T. rex news cool, either).
The point: the content camouflages the intensity.
A boy obsessed with train schedules signals potential autism to most observers. A girl obsessed with horse breeds or anime characters (sorry girls, I’m less in the know with this one) signals normal girl development unless someone examines the depth, exclusivity and functional role of the interest rather than just its topic.
The study suggests diagnostic tools could benefit from adding more female-like examples to prevent misdiagnosis through content-based dismissal of interests that meet all structural criteria for restricted, repetitive patterns.
But the deeper issue is diagnostic criteria designed around male presentation assuming visible, socially disruptive repetitive behaviours as necessary autism markers. Girls demonstrating fewer observable stereotyped behaviours aren’t necessarily less autistic — but they may be more socially aware of which behaviours mark them as different, and could be more motivated to suppress visible difference.
The internal experience — the need for routine, the distress when patterns break, the regulatory function of special interests — is real, and persists regardless of whether external observers can count stereotyped behaviours during assessment, and in spite of whatever parents think.
Parental expectations as a differential assessment standard
The researchers identified the parent-versus-testing divergence as potentially reflecting bias in parental reporting — theorising that parents expect greater cognitive and emotional regulation from daughters than sons, which leads to more critical assessments of girls.
This interpretation matters because it suggests the executive function deficits parents observe in autistic girls aren’t imaginary or exaggerated. They’re real. But they’re being measured against a different standard than equivalent deficits in boys.
A son who struggles with task initiation, can’t shift between activities smoothly, and has emotional regulation difficulties might be seen as “typical boy behaviour” or “ADHD traits” that don’t substantially concern parents until they become severely disruptive, or they can’t seem to direct or change them.
But a daughter demonstrating identical struggles receives more critical evaluation because she’s violating gendered expectations about female emotional intelligence, social awareness and self-regulation capacity.
The same behaviour > different assessment > different likelihood of clinical referral.
The study documented this through BRIEF-2 composite scores. The results: girls showed significantly worse deficits in emotion regulation and cognitive regulation according to parent ratings. These are precisely the domains where gendered expectations operate most forcefully (“girls should be better at managing emotions, should be more organised, and should need less external structure for task completion”).
When autistic girls demonstrate deficits in these domains, parents notice and rate them as severe because they violate expectations. When autistic boys demonstrate equivalent deficits, they may register as less concerning because they align with existing assumptions about male cognitive and emotional development.
The study included only children without intellectual disability who possessed fluent speech adequate for assessment. This means all participants had cognitive capacity for compensation strategies. Yet girls demonstrated fewer observable repetitive behaviours whilst parents reported worse executive function.
The pattern suggests girls are successfully suppressing visible autism markers during assessment through compensation strategies that exhaust cognitive resources, producing executive function collapse observable to parents in daily life but not to clinicians in structured assessment.
And that collapse receives harsher evaluation from parents applying gendered standards about appropriate female cognitive and emotional regulation in home-life.
The female autism diagnostic gap operates through multiple mechanisms simultaneously: assessment tools developed on male populations; gendered interests dismissed as typical; compensation rendering internal struggle invisible during clinical observation; and parental expectations creating differential standards for evaluating equivalent deficits.
The result? Girls who are autistic but don’t meet male-normed criteria, whose interests camouflage rather than signal difference, who pass clinical testing through compensatory effort whilst struggling in daily life, and whose struggles receive harsher assessment from parents expecting more.
To be clear: nothing here is about individual parental failure (thinking so or worrying so is another form of collapse). Instead, it’s another example of a structural bias from the diagnostic gatekeeping system, operating through cultural, gendered developmental expectations that the neurodivergent treatment system doesn’t account for and potentially amplifies.
