# Study 000B Audit

## Structural audit

- Pass: the package is self-contained and reads only from bundled `source_tables`.
- Pass: the source manifest covers `3` packaged input tables.
- Pass: the package includes manuscript, methods, results, discussion, limitations, figures, outputs, appendix, and audit report.

## Scientific audit

- Pass: the study clearly follows `Study 000A` while standing as a second primary program study focused on internal cost and efficiency.
- Pass: the central mixed finding is data-backed. `speed_per_hr_mean` changed `13.41%`.
- Pass: the persistent-burden claim is constrained and explicit. Mean unexplained HR burden shifted from `-1.60` to `3.09` bpm.
- Pass: the descriptive HR model is documented rather than hidden, with coefficients recorded in the package.
- Pass: the late-window interpretation is limited by partial HRV, partial food coverage, and unmatched window span.

## Residual cautions

- The HR model is descriptive rather than physiologically validated.
- The early and late windows are not matched in elapsed time.
- Specialization correlations are likely partly time-confounded.
- Food context remains partial and cannot adjudicate the main finding.

## Audit verdict

`Study 000B` is structurally sound and scientifically coherent. It supports a valuable secondary conclusion for the program: specialization coincided with better adaptive efficiency and more favorable recovery markers, but did not fully remove session-level unexplained HR burden. That mixed pattern is well supported and appropriately bounded.