ECR-000001 Analysis Plan

Layer 1: Packet observations.

Layer 2: Cross-model comparison.

Layer 3: Hypothesis updates by researcher or research engine.

No hypothesis may be updated directly from a single model response. Hypothesis updates require aggregation or explicit researcher review.

• Test whether the instrument can collect structural, primitive, constraint, and representation layers without routine failure.
• Compare whether Scientific Method, Scrum, and Bloom separate in the expected direction.
• Track valid JSON rate, schema compliance, recognition behavior, and unsupported structure.
• Review whether any schema section appears redundant or confusing.
• Do not use calibration alone to strengthen broader theory claims.