# Study 000D Methods

## Study Role

This study is a program-level adjudication study.
It was conducted after the flagship studies and corrective audit were completed.

It does not attempt to rebuild the raw Garmin archive.
Instead, it uses the completed study ecosystem as its evidence base and asks which of two competing adaptation signatures better fits the corrected program.

## Source Packages

The packaged source tables were copied into `source_tables/` from:

- `Study 000A`
- `Study 000B`
- `Microstudy B`
- `Study 000C`

## Competing Adaptation Signatures

### Broad normalization

Expected features:

- broader successful expression across environments
- less dependence on stabilized running context
- more even distribution of gain across cadence and stride
- preservation of stride expression under higher-demand probes
- disappearance of unexplained burden

### Selective expression

Expected features:

- concentration of successful expression in selected conditions
- greater dependence on stabilized running context
- gain assembled mainly through the most accessible lever
- preserved turnover with suppressed stride expression under higher-demand probes
- efficiency improvement without full disappearance of burden

## Evidence Families

Six evidence families were evaluated:

1. ecological breadth
2. mechanical gain pathway
3. higher-demand probe behavior
4. constraint persistence
5. internal cost
6. robustness after ecological correction

## Decision Rule

No hidden score was used.

Each evidence family was evaluated directly and assigned to the signature it matched more closely.
The study answer was then based on the combined pattern rather than on any one metric.

## Important Design Choice

This study is intentionally broader than a metric-ranking paper.
Its task is not to crown one variable.
Its task is to determine which adaptation model the corrected program actually supports.