Bio 227 A&P II at an open-enrollment Hispanic-Serving Institution. Same instructor taught the same course with and without the platform one year apart. 20 matched quizzes. 481 student reflections. Not seeded. Not simulated. Not yet published. This is preliminary evidence that the platform-in-development is producing measurable effects on student outcomes — the kind of signal that justifies committing to the Fall 2026 formal study under an approved IRB protocol.
Q17 covers respiratory structure and function — ventilation mechanics, spirometry, V/Q relationships. Baseline: 78.7% with 7 of 30 students below 70. Pilot: 94.3% with 1 of 28 below 70. A +15.6 point gain and a near-elimination of failure on the single chapter with the most mature platform deployment (v16.1 Section Synthesis, direction-inversion intervention cluster, and 121 student reflections — the largest reflection corpus of any chapter). The respiratory chapter is the pilot's strongest single signal and the pre-registration anchor for the Fall 2026 formal study.
The Bio 227 pilot signal concentrates on reasoning-structured content — physiology, mechanism, cascade, feedback loops, cause and effect. Every substantial gain is on content with an embedded normal-function → failure-mode → intervention structure. The Bio 226 result concentrates losses on pure anatomical identification — bone markings, muscle names, structural nomenclature. The scaffold is a diagnostic reasoning tool; it exploits content structure it wasn't built to substitute for recognition memory. This is not a failure of the scaffold — it is a boundary condition, and a stronger finding than a uniform result would be because it characterizes exactly where the platform applies and where a complementary tool is needed. Reporting both courses honestly is how pilot data earns a prospective study.
The platform's measurement layer evolved mid-semester (March 2026) from per-student in-class checkpoints to post-session transcript audit feeding a Post-Session Announcement workflow. Chapter-level quiz outcomes under the new measurement architecture were equal to or greater than outcomes under the prior architecture — the Spring 2026 digestive chapter (Q19–Q22) averaged +6.0 points over the Spring 2025 baseline on the newer instrument. The pilot is not a fixed-dose intervention; the platform is in active development, and the evidence tracks its iteration. Fall 2026 formal data collection runs under a single locked platform version.
Students who wrote "I used to think X but actually Y" — identifying AND correcting their own misconception — scored 96.7% on corresponding quiz items. The act of articulating the error in writing appears to consolidate the correction. This is the teach-back method measured at scale.
| Chapter | SG In-Class | QuizPrep | Quiz | QP→Quiz Δ | Pattern |
| 227 Ch22 Respiratory | 75.9% | 85.5% | 88.7% | +3.3 | PIPELINE WORKING |
| 227 Ch23 Digestive | — | 92.1% | 91.8% | -0.3 | CALIBRATED |
| 226 Ch05 Integumentary | — | 91.4% | 91.2% | -0.2 | CALIBRATED |
| 226 Ch06 Bones | — | 86.4% | 86.4% | 0.0 | CALIBRATED |
| 227 Ch19 Vessels | — | 91.6% | 88.5% | -3.1 | MONITOR |
| 226 Ch07 Skeleton | — | 89.5% | 82.6% | -6.9 | QP → QUIZ DROP |
| 226 Ch10 Muscular | — | 80.2% | 72.1% | -8.1 | QP → QUIZ DROP |
| 227 Ch18 Cardiovascular | — | 92.5% | 83.0% | -9.5 | TOO SCAFFOLDED |
The QuizPrep-to-quiz deltas cluster by content type, not by chapter. Calibrated or gaining: Ch22 Respiratory, Ch23 Digestive, Ch05 Integumentary, Ch06 Bones functional content — all chapters with embedded reasoning structure. Dropping: Ch10 Muscular (−8.1), Ch18 Cardiovascular structural quiz (−9.5), Ch07 Skeleton (−6.9) — all chapters dominated by pure anatomical identification where the quiz tests recall of nomenclature the practice item couldn't simulate. The pattern confirms what the cross-semester pilot signal already showed: the scaffold produces gains on reasoning-structured content and does not substitute for memorization of nomenclature. The "SG In-Class" column only carries Ch22 data because that chapter was the last one measured under the platform's original per-student checkpoint architecture — the measurement migrated to a post-session transcript-audit workflow in mid-March, and the digestive chapter's gains were captured under the new architecture.
The dental demo shows what the component library builds. The education page shows what the instructional platform is producing during its pilot phase. Same components. Different content. Measurable pilot-stage outcomes. The formal study begins Fall 2026; the platform build continues now.
Start your conversation →