What autistic job
satisfaction data
reveals — the
flatness at the
heart of autistic
employment

A new study measures job satisfaction in 122 autistic employees and finds something it does not name

A study published in Research in Neurodiversity in 2026 by Barbara D’Entremont at the University of New Brunswick set out to identify what predicts job satisfaction among autistic employees. The design is conventional and competent: 122 self-reported autistic adults with autism-spectrum quotient scores of 32 or above, an online survey, hierarchical regression using Social Cognitive Career Theory as the framework, three blocks of predictors entered in sequence — individual factors, work conditions, and employment self-efficacy — against job satisfaction as the outcome. The author’s framing is constructive throughout: identify the modifiable predictors, recommend the workplace adjustments, hand the employer a list of things to do.

The predictors that survived the final model are the unsurprising ones. Perceived organisational support, person-environment fit, and employment self-efficacy each contributed significantly. Positive affect washed out once work conditions were entered. Sensory sensitivity, empathy, communication scores, and sex assigned at birth were not associated with satisfaction at all.

The author lands, funnily enough, where the methodology directs: employers should foster supportive environments, ensure alignment between worker and role, and build self-efficacy through mastery experiences. Reasonable, evidence-based, and employer-facing. So, the piece you’d expect.

The finding the paper does not pause on is the one that should organise everything that follows. Across the full sample of 122 autistic employees, mean job satisfaction was 3.06 on a 5-point Likert scale. Three. Point. Oh. Six. The midpoint of the scale corresponds, in the instrument’s own wording (!), to “undecided.” Not satisfied. Not dissatisfied. Neither. The definition of meh. The paper notes this in a single sentence — “participants were ‘undecided’ about their job satisfaction” — and moves directly to the predictor analysis, treating the mean as descriptive context rather than as a finding requiring interrogation.

That mean is the finding. Everything else in the paper is a mere footnote to it, in my opinion.

21.7% of autistic individuals in employment, and the most fortunate of those still cannot endorse satisfaction

To understand what the 3.06 actually represents, the sample has to be situated. Three-quarters of the participants had completed some level of post-secondary education. Nearly four in five were currently employed at the time of the study, the majority in competitive rather than supported or sheltered roles. They were predominantly from the United States, Canada, and the United Kingdom — countries with comparatively developed disability-employment infrastructure — and they had found their way to an online survey hosted across autism communities and message boards. This is, by every demographic marker the paper records, the most functional and most fortunate end of the autistic workforce.

That qualifier matters because the surrounding context is brutal. UK Office for National Statistics data put autistic employment at 21.7%, roughly one-quarter the rate of non-disabled adults. Department for Work and Pensions figures show 31% of people with a neurodiversity condition in employment compared to 54.7% of disabled people overall — meaning autistic adults are the least likely group, even within the disabled population, to be working. World Economic Forum estimates place US autistic unemployment at 85%. Behind the 122 participants in this study sits a funnel of attrition: most autistic adults are not in the workforce at all. Most who do get hired do not stay. Most who would have answered a job-satisfaction survey did not get one, because they were not in a job to evaluate.

The 122 in the D’Entremont sample are, in other words, the survivors of that funnel. They are the ones who got through the hiring gate, stayed long enough to be employed at the moment of the survey or to remember employment recently enough to report on it, and were functional enough to find and complete an online questionnaire.

The mean they produced — 3.06, undecided — describes the affective state of the autistic workforce at its most fortunate. What the data look like across the missing 78.3% is not directly measurable here. What can be said is that the mean job satisfaction reported by the visible quarter is not satisfaction. It is flatness.

Flatness is the right word, because it forces a distinction the field has been sloppy about. Not-suffering is not thriving. An autistic worker who can answer “neither agree nor disagree” to whether they like their job has not located satisfaction — they have located the absence of strong dissatisfaction alongside the absence of any real satisfaction, which is a different state and worth a different name. Surviving employment and the workplace is not the same as flourishing within it, and the data here are evidence that even the autistic workers least likely to be ground out of the labour market entirely are stranded in survival mode rather than reaching anything beyond it.

What the predictors actually measure when you read them the other way around

The paper’s framing of the predictors is forward-facing and employer-friendly: perceived organisational support, person-environment fit, and self-efficacy are modifiable, so employers can move them, so satisfaction can be improved. Take the same data and reverse the read. The predictors are not just levers for change. They are the markers of how rare the conditions under which autistic workers actually report satisfaction are.

Perceived organisational support correlates with satisfaction because most autistic workers do not feel their organisation values their contributions or cares about their wellbeing, and the few who do feel it report better experiences. Person-environment fit correlates with satisfaction because most autistic workers occupy roles that do not align with their skills, education, or values, and the rare ones in roles that do report better experiences. Employment self-efficacy correlates with satisfaction because most autistic workers have accumulated employment histories of misfit, rejection, masking, and exhaustion that have done the opposite of building confidence, and the few who managed mastery experiences anyway report better experiences.

Read forward: the three predictors are a list of employer to-dos. Read sideways: they are a measure of how exceptional the conditions for autistic job satisfaction actually are.

The author is honest about which direction the modifications run — work conditions are “more readily adaptable than some individual factors” — but the implicit assumption is that the conditions are absent for individual organisational reasons rather than structural ones. The data are consistent with a different reading: that workplaces designed end to end for the neurotypical default produce, by their architecture, low support and low fit and low self-efficacy as the norm for autistic workers, and that the satisfaction differential measures the rarity of environments that happen to break that pattern. Same numbers. Different lever.

This is the workplace-is-the-variable argument the Directory has made before, arriving as job-satisfaction data this time. The square peg has not changed. What changes the reported satisfaction is the rare hole that happens to fit it.

The inside view of working below the critical neurodivergent mass threshold

The D’Entremont study reads differently — and lands harder — when placed alongside the van Rijswijk team-composition work covered earlier this month. That study, looking at 357 employees across 70 teams, found a within/between contradiction the field had been collapsing into one finding. At the individual level, the isolated neurodivergent worker reports lower psychological safety than their neurotypical colleagues, and the deficit drags down team identification and relationship quality. At the team level, the relationship reverses: teams with a higher proportion of neurodivergent members report higher psychological safety, identification, and relationship quality than mostly-neurotypical teams. The lone neurodivergent employee suffers. The neurodivergent-dense team thrives.

The two studies describe the same condition from opposite ends. Van Rijswijk shows when neurodivergent workers do well — when there are enough of them in one place to reset the team’s norms. D’Entremont, sampling almost entirely autistic workers in conventional workplaces where they were almost certainly the only autistic person on their team or one of very few, shows what most autistic workers actually feel. The 3.06 mean is the affective endpoint of being the sample-size-of-one in environments calibrated for someone else, year after year. It is what working below the critical neurodivergent mass threshold looks like, measured at the level of subjective experience rather than at the level of group dynamics.

What the two pieces together suggest is that the autistic workers who reported flatness in the D’Entremont sample were not, in the main, suffering from poor employers or poor fit in some individually-correctable way. They were occupying positions in which the structural conditions for satisfaction — the critical density that produces adaptive norm creation, distributed masking load, explicit communication norms, decentralised support — were absent by default. The few exceptions in that sample, the ones whose perceived organisational support and person-environment fit and self-efficacy scores were high, were the ones who had stumbled into the rare configuration where some of those conditions happened to be present anyway, usually through individual employer attention rather than actual structural design. Most autistic workers do not stumble into that. They stay isolated, and they report flat.

The honest reading of the D’Entremont data is therefore harder than the paper’s own framing allows. Employer-level adjustments help — that much the data support. But the larger story is structural, and the flatness in the survivor sample is not a problem the existing workplace can fix by trying harder at its current design. It is what the existing workplace produces, by design, in autistic workers who are doing everything else right. The 21.7% who made it through the gate are not thriving. They are not failing either. They are stranded at neither, because neither is what the architecture they entered was built to produce in workers whose architecture does not match it.

A workforce that can only locate the absence of dissatisfaction is not the success the disability-employment statistics imply when employment rates tick upward. Getting more autistic people into work without changing what work is would, on these numbers, mean moving more people from outside the funnel into the flatness. That is a different policy ambition than the one the field has been talking about, and it is the one the data here suggest needs to be talked about next.

Citations

D’Entremont, B. (2026) — Predictors of job satisfaction of autistic employees

van Rijswijk, J., Curșeu, P. L., & van Oortmerssen, L. A. (2026) — A Relational Perspective on Neurodiverse Teams: Psychological Safety, Relationship Quality, and Identification

The Neurodiversity Directory — Neurodiversity Statistics (2026)