Report status

The PDFs retain their bytes from release baseline a369ee0dc2dcac24feaa9e2e1faf160c39d5a3de, which revised the thesis and poster. Locations use PDF pages, with printed page numbers where they differ. Use the current code and tables for reproducible calculations; the source gaps below limit interpretation of the historical reports.

Report Status
Technical report, 12 pages Historical methods and results
Thesis, 53 pages Protocol, analyses and consent appendices
Poster, 1 page Proposed protocol and expected results

Corrections and source gaps

Location Correction or unresolved issue
Technical p3 The report counts openness values above 1 as closed; the code counts values at or below 1 as closed. Units and polarity are unverified. Current outputs are heuristic candidates that require observed reopening and handle gaps. Gaze distance was not angular velocity.
Technical p6 P01 S3 interval SD is 1012.69 in the report and 65.39 in the CSV. The original raw source is unavailable. Both values are preserved; the discrepancy is unresolved.
Technical p8 to 11 Claims about agreement at k = 3, PCA before clustering, HADS validation and stable traits exceed the code evidence. The case study clusters timestamp samples from one person. It does not validate clinical or participant categories.
Thesis p25, printed 28 Some p values exceed 1; values shown as 0.00 are rounded. Recovering lost exponents and precision requires the original statistical output. No replacement values are supplied.
Thesis p37 to 38, printed 40 to 41 Available eye and cardiac analyses differ from the planned GSR and facial integration. Diagnostic and stable-trait conclusions exceed the evidence from the uncertain ten-person estimates.
Thesis p4 to 5, printed 7 to 8 The protocol limits access to the research team. It differs from the author's later account of consent for sharing. No review identifier is independently established.
Thesis p46, printed 49 Annexes E/F explicitly record photo consent.
Poster The poster gives a target of 10 participants, expected results and a planned CER submission. Its Tobii/Empatica equipment, PHQ/ACEs instruments and age range of 18 to 30 differ from later methods. It does not document approval or measured results.

Cardiac sensitivity

The original P01 S3 export is absent from the released files and Git history. The sensitivity analysis compares the released value, the report's printed value and omission of P01 from all three group tables.

Scenario Participants ICC(1,1) of interval-sample SD 95% interval
Released CSV: P01 S3 = 65.3938 10 0.221 −0.131 to 0.659
Report candidate: P01 S3 = 1012.69 10 0.038 −0.253 to 0.509
Omit P01 from all metrics 9 0.216 −0.150 to 0.680

Cardiac ICC intervals and session correlations for the three sensitivity scenarios

The report value is rounded and unverified. Neither alternative establishes the correct value. Released tables remain unchanged. Substitution changes eight of 36 Spearman correlations, with a maximum absolute change of 0.109. Pupil and response-duration ICCs remain unchanged in that scenario. Participant omission changes all three metrics' estimates.

ICC results · Correlations · Omnibus tests · Paired tests · Sources and settings

python -m scripts.build_group_sensitivity --check
python -m scripts.build_group_sensitivity --output-root /tmp/mms-sensitivity

Bibliography

The Green/Harlow eye-tracking attribution to Journal of Anxiety Disorders 73:102233 is incorrect. That identifier belongs to McKay, Yang, Elhai and Asmundson's 2020 study of COVID-related disgust and anxiety, DOI 10.1016/j.janxdis.2020.102233. It does not support the attributed eye-tracking claim. The historical Johnson/Bennett titles remain unverified.

Current evidence

Data contract · Figures