Parents rank ABA
last on ethics —
and the ones
whose children
had it are the
least critical of all

The numbers — a gap that isn’t close

Eighty-seven parents of autistic children across Canada answered a survey about the three services their children most commonly encounter: occupational therapy, speech-language pathology, and applied behaviour analysis. The team — Isabelle Caven and colleagues at the University of Toronto and Holland Bloorview, with an autistic advisor as co-author and an advisory group of autistic people, parents and clinicians — scored acceptability out of a possible 64.

Speech-language pathology scored a median of 41.5. Occupational therapy, 40. ABA, 28.

The gap widens where it matters most. Asked whether each therapy’s goals aligned with their personal values, around 78% of parents agreed for OT and 86% for SLP. For ABA, 31% agreed — and 39% strongly disagreed. On methods, the pattern repeats: roughly three-quarters agreement for OT and SLP, under 30% for ABA, with just over 40% strongly disagreeing. This was not a survey of autistic adults, whose accounts of behavioural intervention are by now well documented. These were parents — the people who select, fund and sit in the waiting rooms of these services — and a plurality of them rejected ABA’s goals and methods at the level of values.

What they objected to is specific. Free-text responses described the potential harms of training children toward neurotypical appearance: disempowerment, removal of self-soothing mechanisms, loss of a sense of self, and compliance elevated to a benchmark of success. One parent’s account of ABA’s purpose was that it exists to train socially acceptable behaviour so that other people are not made uncomfortable. That is the same structural observation this corpus has made about play research that codes “improvement” as playing less like yourself, about why stimming reads as executive deficit to instruments built for compliance, and about the logical impossibility of normalising the autistic brain — except here it is arriving unprompted from parents, in a survey box.

And that word, compliance, is doing more work than anyone in the exchange quite acknowledges. There is a pattern that recurs across this whole field once you look for it: a system measures compliance and calls it effort; a manual measures compliance and calls it disorder; a classroom measures compliance and calls it character. What these parents have caught is the therapeutic instance of the same move. A child who sits where they are put, responds when prompted, and stims less than they did is producing compliance behaviours — and the question of whether compliance is progress cannot be answered by the instrument that records it, because the instrument was built to count exactly that. The parents were not disputing the measurements. They were disputing what the measurements were measurements of.

What “acceptable” was allowed to mean

The instrument matters as much as the result, because for once the instrument was not built by the intervention it was measuring. Behaviour analysis has long audited itself through social validity — Montrose Wolf’s 1978 construct asking whether an intervention’s goals, procedures and outcomes are perceived as acceptable. The authors note, citing a forty-year scoping review, that social validity assessment has been criticised for a lack of rigour. It is the field marking its own homework, and marking it generously.

Caven and colleagues instead used the Theoretical Framework of Acceptability from the health-services literature — seven constructs including Ethicality, defined as fit with an individual’s value system. And the finding with the longest shadow is which constructs parents actually engaged with. Given free text, they wrote richly about three: Ethicality, Perceived Effectiveness, and Self-Efficacy. Ethics came first, unprompted, across all three therapy types. Parents wanted services that draw on a child’s strengths, honour multiple communication forms, and do not treat the reduction of visible autistic traits as an outcome — a set of demands that maps closely onto what quality-of-life research keeps finding matters and what the research on autistic joy shows was simply never measured. Change the acceptability instrument, and the verdict changes with it. That is not a flaw in this study. It is the study’s finding.

It also lands amid a wider pattern this corpus has tracked: intervention-before-five neuroplasticity arguments used to justify intensity, school programmes running on evidence bases of five small studies, and a sector that increasingly renames its practices without redesigning them. Parents, on this evidence, have noticed.

The finding that cuts the other way

Now the part an advocacy piece would omit. Parents whose children had actually accessed ABA rated it more acceptable — median 30 — than parents whose children had not, median 25.5. The difference is statistically significant (t = 3.36, p < 0.01), and it appeared for ABA alone; for OT and SLP, experience made no difference. Some experienced parents credited ABA with faster visible progress than the alternatives; one described comparatively rapid goal attainment after slow results elsewhere, while acknowledging it would not hold for every child.

This corpus has a stake here, so let it be stated rather than smuggled: The Neurodiversity Directory does not list behaviour-analytic provision, and its verification standard is built on the paradigm those excluded services sit outside. A finding that experience of ABA raises its acceptability therefore deserves the most careful reading available, not the most convenient one. Three readings fit the data, and the study cannot separate them. Selection: parents whose values already accommodated behavioural approaches were the ones who enrolled their children, so the causal arrow may run from acceptance to access, not access to acceptance. Information asymmetry: the authors themselves suggest non-experienced parents may be informed by external sources — prominently, autistic adults’ published accounts of harm — so the gap may measure exposure to testimony rather than exposure to therapy. Genuine revision: some families observed benefits and updated. Nothing in a cross-sectional survey adjudicates between these — and honesty requires adding that the third reading, even if true for some families, speaks to Perceived Effectiveness, which was never the ground on which the objection stood. A therapy can produce visible compliance efficiently and still fail on Ethicality; efficient production of compliance is, on the critics’ account, precisely the problem. The parents’ own top-rated construct keeps the two questions apart even when the headline number blurs them.

There is a further wrinkle worth reporting straight: parents across the sample — including critics — identified reducing behaviours that cause harm, like self-injury or elopement, as legitimate territory, and the authors note that behavioural principles operate informally everywhere from classrooms to reward charts. The word “behavioural” is carrying too many meanings for one survey to weigh, and the authors call for exactly that specificity in future work.

What this survey can and cannot carry

The limits, plainly. Eighty-seven self-selected respondents recruited through autism organisations and social media, skewed toward higher income and education, lacking racial and ethnic diversity by the authors’ own account, with 88.5% women — the gendered division of who answers for a child’s care is itself data — and around 14% of respondents autistic themselves. Data were collected in 2022 and published in August 2026; a four-year lag in a fast-moving discourse means today’s parents may sit further along the same trajectory, or not. Fewer than half the sample had direct ABA experience. Senior author Melanie Penner declares consulting fees and research grants including from Autism Speaks — a funder much of the autistic community regards as adversarial — which, whatever one concludes, makes the ABA-critical result harder to dismiss as activist capture. And the ABA umbrella covered everything from discrete trial training to parent consultation, unanalysed by type.

One absence deserves its own sentence. The study was designed as part of a larger project comparing parents, clinicians and autistic people — and the authors report that despite their efforts, they could not reach enough autistic respondents to compare against the parent sample. The constituency whose acceptability judgement matters most is the one the survey infrastructure could not reach, in a project with autistic advisors and accessible design. That is not this team’s failure so much as the field’s standing condition, and this corpus has argued the consent and participation problem in autism research runs exactly this deep: the systems for gathering opinion were built along the same channels as the systems for delivering therapy, and they miss the same people.

What survives the limits is the structural point. For decades the acceptability of autism intervention was assessed with instruments the intervention’s own field designed, answered mostly by proxies, and scored against clinician-defined outcomes. Move the frame one step — ask parents about values, with a framework borrowed from outside the discipline — and the flagship intervention drops thirteen points below an occupational therapist teaching a child to manage their own morning.

And look at what the parents in the high-scoring conditions were actually describing, because it is not therapy in any specialised sense. Multiple forms of communication accepted. Environmental triggers identified and removed. The child’s own preferences treated as information. Predictability where predictability helps. None of that is a clinical intervention; it is calibration that would serve any child in any room, which is precisely why it scores well on ethics — there is no verdict inside it. That it reaches families through an occupational therapy referral, funded by a diagnosis, is the usual arrangement in this field: universal design sold first to the licensed few, admission charged at a members’ door to a room that was always everyone’s.

So the demand this generation of parents is making, on this evidence, is not for better-behaved children. It is for services whose definition of improvement their children could endorse — a shared definition of the goal, set with autistic people rather than for them. Any therapy that cannot survive that renegotiation was never measuring the child but the child’s distance from an average that describes no actual person, and handing the distance back to the family as progress.

Citations

Caven, I., Hamdani, Y., Lunsky, Y., Weaver, C. & Penner, M. (2026) — Parent perspectives on services available to autistic children in Canada: A mixed methods survey study — PLOS One

Wolf, M. M. (1978) — Social validity: the case for subjective measurement or how applied behavior analysis is finding its heart — Journal of Applied Behavior Analysis

Anderson, L. K. (2022) — Autistic experiences of applied behavior analysis — Autism