Contexts: class size, multi-grade, and facilitator realities

The curriculum must work not only in a well-resourced classroom but in the messy, varied contexts where most learners actually are. This document states what the evidence supports and how the curriculum adapts. Grounded in the vetted sources in ../sources/sources.yaml.

The evidence, briefly

One-to-one tutoring is the ceiling. Bloom’s “2 sigma” work (1984) found tutored students scored about two standard deviations above classroom-taught peers. The practical target is to get as close to tutoring as possible without needing a tutor per child. The biggest verified, scalable levers from that same work: mastery learning / feedback (≈1.0 SD), classroom participation (≈1.0), and cooperative learning (≈0.8).
Relationships and feedback beat class size. Meta-analyses (Hattie, Visible Learning) rank teacher–student relationships and feedback among the strongest influences, and class size among the weakest.
Class size is real but narrow and contested. Project STAR (randomized, K–3, small 13–17 vs regular 22–25) found gains — about doubled for minority and low-income learners — but recent meta-analyses find the effect negligible outside STAR-like designs and classes under ~15. Practical reading: the youngest and most disadvantaged benefit most from smaller groups, but group size is far less important than what happens inside the group.
Teach at the learner’s level, not their age. Pratham’s Teaching at the Right Level (group by assessed level, not grade) produced large, cheap gains in literacy/numeracy across randomized evaluations (J-PAL) and is a “Good Buy” (GEEAP/World Bank).
Multi-grade, self-paced, low-tech schooling works. Escuela Nueva (Colombia, 1970s–) runs one-room multi-grade rural schools on self-paced learning guides and student cooperation — ~20,000 schools in Colombia, adopted in ~19 countries, and a UNESCO-noted case where rural schools outperformed urban ones.

How the curriculum adapts

Class size

We cannot control class size; we can design so the ratio-to-relationship is favorable at any size:

Small-group and pair structures are the default, not whole-class lecture.
Peer teaching — learners who have mastered a step help those still learning it (the Escuela Nueva pattern).
Self-assessment and portfolios carry the load that a single teacher cannot observe individually.
Whole-class checkpoints are brief; the heavy lifting is done in small groups and independently.

Multi-grade / multi-age

Because the curriculum is a spiral, a single room of mixed ages can work on the same domain at different depths. When a teacher has several grades at once:

Use self-paced guides (Escuela Nueva style): each learner works through their own level while the teacher rotates.
Group by mastery level, not age (TaRL), and regroup as needed.
Pair older/more-advanced learners as peer guides for younger ones.

No facilitator / minimal facilitator

Lessons are written to the learner, so a self-directed learner can follow the ## Procedure alone.
For groups without an adult, use self-organized learning around big questions (Mitra’s SOLE): pose the essential question, let small groups investigate, and share back. Note: the “Hole in the Wall” claims are contested — treat SOLE as a useful structure, not a miracle.
The ## Facilitator note degrades gracefully: a learner or older peer can act as the facilitator.

Cultural power dynamics

The curriculum’s “question the answer” stance assumes it is safe to question adults. Where that is not the case, the facilitator should:

Frame questioning as wondering together, not challenging authority.
Use written or anonymous questions (the “wonder box”) so learners can ask without speaking up.
Lead by example — the facilitator wonders aloud too.

When to instruct, when to inquire

Foundational skills (reading, arithmetic, novice content) are taught with explicit instruction, worked examples, and guided practice — the evidence is clear that minimal guidance fails novices (Kirschner, Sweller & Clark, 2006). Inquiry and story then build on those foundations for concepts and motivation. We hold both, and we make neither a dogma.