Camouflaging research was measuring two different things

Our first masking note examines the 2017 interviews that gave the field much of its language. This 2021 review asks what happened when researchers tried to turn that language into numbers.

That is an important second step. “Masking” can mean hiding a movement, rehearsing a script, changing an expression, copying a peer, or producing a social performance that looks easier than it feels. If studies measure different parts of that process, combining their conclusions without care can make the evidence sound more settled than it is.

The reviewers asked five broad questions: how camouflaging had been measured; whether it differed by autistic traits or diagnosis; whether it differed by sex or gender; whether it changed with age; and how it related to mental health.

They also examined who had actually been included. A finding about “autistic people” carries a different weight when samples are mostly white, verbally fluent, diagnosed adults recruited through online networks.

The team registered a protocol, searched five research databases and three grey-literature sources, checked references, contacted authors, and updated the search in May 2021. Two reviewers assessed full texts. Four authors split data extraction and cross-checking, and they used the Mixed Methods Appraisal Tool to assess each study’s design.

Twenty-nine studies met the criteria: 18 involving adults and 11 involving children or adolescents. After overlapping datasets were counted once, the review described 2,254 adults and 1,077 younger participants. Adult samples were mostly women, usually formally diagnosed, and often diagnosed in adulthood. Among studies that reported race, most participants were white. IQ and co-occurring conditions were often missing.

The measures fell into two broad families. “Internal-external discrepancy” methods calculated a gap between an estimate of inner autistic traits or social-cognitive performance and an observer’s rating of visible social behaviour. Self-report methods asked people about their own strategies, effort, frequency, or contexts.

Three preliminary patterns appeared.

First, adults reporting more autistic traits tended to report more camouflaging. Having an autism diagnosis also travelled with more self-reported camouflaging in some studies, though the review suggested the underlying traits may matter more than the label itself.

Second, most studies comparing sex or gender found higher average camouflaging among females or girls and women. The rest generally found no difference; none found the reverse. But effects were often small to moderate, methods varied, and camouflaging was also documented among males and boys or men. Samples of non-binary people were too small for a stable conclusion.

Third, higher self-reported camouflaging among autistic adults was associated with more mental-health difficulties. The same relationship was not consistently found when camouflaging was defined as an observer-rated discrepancy. That difference is revealing: feeling a need to manage one’s social presentation may not be the same thing as how successfully an observer thinks it is hidden.

Not one validated construct. The two measurement families may capture connected but distinct parts of camouflaging. Even within each family, studies used different scales, tasks, and scoring decisions. The reviewers could not pool the results in a meta-analysis.

Not causality. Most mental-health evidence was cross-sectional. It cannot show that camouflaging caused anxiety or depression. Distress could increase camouflaging, or stigma and lack of acceptance could contribute to both.

Not a representative community sample. Twenty-three studies did not recruit a representative autistic sample. Reporting about race, gender identity, intellectual disability, communication, and participant flow was thin. Adult men, non-white people, non-binary people, people with intellectual disability, and adults diagnosed in childhood were underrepresented.

Not a women-only phenomenon. Average group differences do not make camouflaging uniquely female or present in every woman. The review explicitly warns against applying its conclusions to the whole autistic community.

The paper’s most useful finding is methodological. A person can look less visibly autistic without feeling that they are deliberately camouflaging. Another person can expend enormous effort while an observer still notices their differences. Calling both outcomes the same thing hides the part that may matter most to wellbeing: the sustained work, pressure, and loss of choice.

That distinction also sharpens everyday language. When someone says they mask, ask what they mean. Are they suppressing movement, monitoring eye contact, borrowing scripts, avoiding disclosure, changing how they speak, or recovering after social performance? A score cannot carry all of that unless the measure says which part it captures.

The review is candid about its own boundary. The authors had published in this field, and the review did not include participatory design or autistic input. Technical care does not replace community perspective when the construct is about a lived strategy.

The adult evidence mostly represents formally diagnosed, verbally fluent adults recruited online or through autism organisations and research databases. Sixty percent of adult participants were reported as female or women; among the eight adult studies reporting race, 86.1% were white.

The child and adolescent literature leaned the other way by sex or gender, with 62.9% reported as male or boys. Both sets had large gaps in reporting. The review gives a map of the evidence available by 2021, not a complete map of camouflaging across autistic lives.

Research should state whether it is measuring strategy, effort, frequency, motivation, visible effect, or some combination. Studies should use more than one method and test whether the methods agree for the same person. Longitudinal work could then ask whether pressure to camouflage predicts later distress, rather than assuming direction from a one-time correlation.

Participatory studies are also essential. Autistic people should help decide which outcomes matter: not merely whether a behaviour becomes less visible, but whether the person has more choice, safety, energy, belonging, and access to support.