# Study 000C Methods

## Program question

After exhausting the corrected full dataset, what survives unchanged, what must be refined, and what stronger final program conclusion is actually supported?

## Inputs

This package reads only from bundled source tables copied from the current canonical study ecosystem and the full-system audit:

- `Study 000A` yearly overview, phase model, mechanics contrast, decomposition, variability rank, and window context
- `Study 000B` window summaries and specialization correlations
- `Microstudy A` surface summary and anchor-context tables
- `Microstudy B` outdoor cases, sign tests, and adjacent-dynamics summaries
- `Study 000AB` bridge findings, evidence ladder, and consistency audit
- `Program Audit` hybrid-cardio candidates, monthly hybrid summary, window-overlap table, and impact matrix
- yearly ecology table from the current Stage 5 main

## Exhaustive checks performed

1. Yearly ecology was recalculated into a corrected ecology summary by adding GPS-bearing hybrid-cardio counts, hours, and miles by year.
2. The early and late flagship windows were re-evaluated against the hybrid-cardio audit layer.
3. The later Microstudy B outdoor probes were checked against the later hybrid-cardio block to determine temporal separation rather than assuming contamination or no contamination.
4. A claim-stability matrix was built to test which major program claims were retained, refined, corrected, or cleared.
5. A final full-program model was then written only after the corrected ecology and separation checks were complete.

## Interpretation rules

This study treats the running-specific mechanics findings as valid unless the exhaustive cross-check reveals direct contradictory source evidence. It treats ecology claims more cautiously, because Garmin label taxonomy can collapse outdoor hybrid conditioning into `indoor_cardio`. As in the earlier packages, all conclusions remain descriptive rather than mechanistic proof.