elaboration-and-self-explanation

Elaboration and Self-Explanation

Overview

Students learn more when they generate the explanation than when they read yours — the self-explanation effect (Chi et al. 1989) and elaborative interrogation (reviewed in Dunlosky et al. 2013). The lever: don't pre-write the "why" — prompt it.

The shift: prompt the why, don't supply it

A reveal that explains the reasoning beats one that gives only the number — but a prompt that makes the student explain the reasoning beats both. Supply your "why" in the check-after reveal so they can verify, but first ask the student to produce it.

How to author it

• Ask "why" / "how do you know," not just "what." "Why must the total energy be negative for a bound orbit?" beats stating "the total energy is negative."
• Predict-then-explain, not just predict-then-check. After a prediction, ask why they expect it — then reveal. The explanation attempt is the learning; the reveal is the feedback.
• Connect to prior knowledge (elaborative interrogation): "How does this relate to the conservation law from last week?" forces an integrating link.
• Self-explanation on worked examples: between steps, ask "why is this step valid?" rather than narrating every justification yourself.
• Then supply your "why" in the gated reveal — as confirmation, not substitute.

Example

• ❌ Supplied: "Notice the mass cancels, because gravity and inertia scale together."
• ✅ Prompted: "The mass cancelled. Before reading on, explain why a heavier block slides exactly as far. (Open the note to check your reasoning.)"

Red flags — STOP

• Every "why" is answered for the student; none is asked of them.
• Predictions with no request to justify the prediction.
• Worked examples that narrate every justification, leaving nothing to self-explain.
• Reflection prompts that ask only "what" (recall) and never "why/how" (elaboration).

Eval-side companion: complements rubric dimensions 1 (retrieval) and 9 (feedback quality) — docs/website/reference/evidence-based-pedagogy-rubric.md. No dedicated scored dimension yet; cite Chi et al. 1989 / Dunlosky et al. 2013 directly.