For decades, special education argued that disabled students need explicit, teacher-directed instruction instead of constructionist learning. The reasoning ran through deficit: language difficulties, processing differences, recall trouble — locate the problem in the student, prescribe the worksheet. In 2025, the field measured its own claim. It does not hold.
A systematic review and meta-analysis in Science Education — National Science Foundation–funded, screened with PRISMA, coded against the Council for Exceptional Children’s quality indicators — synthesized 26 studies of inquiry-based science instruction for disabled students. It is the first review to span disability categories, include both group and single-case designs, and separate the types of inquiry rather than lumping them together. The students learned. And the more autonomy they had, the more they learned.
What the numbers say
Overall effects were large: a combined effect size of 0.79 across group-design studies and a between-case standardized mean difference of 2.76 across single-case studies. But the breakdown is the argument. In group designs, open inquiry — students directing their own questions and experiments — produced an effect size of 1.50. Guided inquiry produced 1.22. Structured inquiry, the teacher-demonstration, explicit-content model long held up as the safe choice for disabled learners, produced 0.39.
Openness won. Not despite the supports, but with them in place.
SWD are capable of engaging in inquiry‐based science instruction and, with appropriate supports, can build the skills necessary for scientific literacy and future success.
Wilson et al., Inquiry‐Based Science Instruction for Students With Disabilities
Inquiry-access is toolbelt theory without the name
The review’s most useful contribution is a category the authors call “inquiry-access” interventions. These are studies where inquiry was present in both the treatment and the control condition, and the intervention was not a different pedagogy — it was added tools. Task analyses. KWHL graphic organizers. Self-monitoring checklists. Peer mediation. Self-directed learning models. The point of these tools was to let disabled students do the authentic scientific work in the authentic environment, rather than being pulled out of it or handed vocabulary drills in its place.
Inquiry-access produced medium effects in group designs and very large effects in single-case designs. That is toolbelt theory arriving under another name. Match the tool to the task and the environment, and the learner stays in the learning. The skill and the support work in tandem — exactly the move we make when we say the tool belongs to the task, not to the diagnosis.
Constructionism, confirmed from inside the fortress
The paper opens by indicting didactic, textbook-driven science teaching and closes by saying inquiry “prepares students to actively construct knowledge about the world around them.” That is the claim of constructionism — people learn by making and doing — now corroborated by the research tradition most historically skeptical of it. Disabled students do not learn science despite doing science. They learn it by doing science.
The review proposes inquiry as a continuum, with supports fading as learners grow more independent and instruction moving toward openness over time. That gradient mirrors the progressive movement we design into Cavendish Space — caves, campfires, and watering holes traversed according to a learner’s interaction capacity, not a fixed sequence imposed from outside. The evidence has reached the door of the room we have been building.
Where it stops
This research converges on the instructional and relational layers — active learning, access supports, an autonomy gradient — and stops short of the layers where learners hold design power.
The tools are chosen and administered by adults, not assembled by learners. Toolbelt theory insists that learners choose their own tools, because tool choice is the skill that travels out of the classroom and into a life. A few of the included studies edge toward this with self-monitoring and self-directed learning models, but tool agency stays mostly in adult hands.
The environment appears only as a coded variable — general education, special education, pull-out — never as a design surface learners can modify. Niche construction means learners directly change the environment to enhance their own chances of success. In this literature, learners move through environments adults have built. They do not build.
And the pathology-paradigm vocabulary persists, even as the paper’s own evidence works against it. Poor science outcomes, the authors concede, “may be associated with how science has historically been taught” — yet the difficulties are still located in students rather than in didactic environments. There is no neurodiversity vocabulary, no social model, no deficit-ideology critique, no participant voice, no co-design. The authors themselves flag the missing demographic data. Nothing about us without us is structurally absent.
This is the recurring shape. Strong research converges toward our framework at the instructional and relational layers, and stops short of the environmental and systemic layers — the layers where disabled and Autistic learners stop being recipients of good design and become its authors. The meta-analysis proves the door is open. It does not yet hand over the keys to the room.
Wilson, S. E., Therrien, W. J., Gersib, J., Rojo, M., VanUitert, V. J., Lovette, G., Longhi, M. A., Benson, S., Powell, S. R., & Doabler, C. T. (2025). Inquiry‐based science instruction for students with disabilities: A systematic and meta‐analytic review. Science Education, 1–15. https://doi.org/10.1002/sce.70029 (Open access, CC BY 4.0)