Amendment for Trajectory-Sensitive Reanalysis of Zniva, Pauli & Schulz (2017): Temporal Granularity of Cardiovascular Recovery Following Social Support

This amendment adds three documents and two code-repositories that together complete the post-registration record for the associated manuscript (Schulz & Graf, ICFT 2026): (1) an OSF amendment documenting all post-freeze deviations and corrections, (2) the online supplement referenced in the manuscript, and (3) a post-review diagnostic requested by a peer reviewer. (1) OSF Amendment. This document (provided in German, the language of the preregistration, and in English translation) fulfils the disclosure commitment made in Section 2 of the preregistration. It documents, without gaps, all code changes between the original OSF freeze (feature lock SHA-256 af1600d9dae181dde4d59a574b88d7d667cea44cee0da0c0d8c8a716905ea795) and the final analysis run underlying the manuscript's results, across two sequential phases: (a) an initial bug-fix log (13 entries, 2026-08-13) correcting programming errors discovered immediately after unblinding, and (b) a structured "Freeze-2" remediation (2026-08-14 to 2026-08-16), triggered by an independent implementation audit, implemented entirely result-blind under a two-stage, hash-bound gate procedure (Gate R / Gate S). No group-dependent computation was permitted until all corrections were complete and had passed an independent delta audit and a cross-implementation audit; the gate sign-offs are recorded with cryptographic hashes binding the feature set, specification, and code manifest used for the final run. No correction in either phase altered a previously inspected group-dependent conclusion; H1 and H2 remained inconclusive throughout. (2) Online Supplement. This document provides the complete numerical results underlying Table II and Sections IV-E/IV-F of the manuscript, including all bootstrap confidence intervals, the full F5 convergence-diagnostics and convergence-sensitivity grids explicitly referenced in the manuscript text, complete results for robustness checks 9.2 and 9.4–9.7, and the complete post-registration (Gate-S) diagnostics (self-computed source–response sensitivity, M40, M46). All figures were verified against the gate-authorized result files and match the manuscript exactly. (3) Post-review confusion-matrix diagnostic. During peer review, a reviewer noted that balanced accuracies for the four preregistered representation levels were below the three-class chance reference and requested inspection of the underlying confusion matrices. This note documents a purely descriptive post-review diagnostic that decomposes the already-realized, frozen leave-one-participant-out predictions into per-class confusion matrices, with an integrity check confirming that the reproduced balanced accuracies match the reported values to numerical tolerance. It performs no model selection, tuning, or additional hypothesis testing, and does not modify the registered H1 or H2 analyses, their decision rules, or their interpretation. The accompanying script and full diagnostic output directory are archived alongside this note. Together, these documents complete the transparent record of the registered analysis, its post-freeze corrections, and its response to peer review, without altering any preregistered hypothesis, decision rule, or conclusion.