Supplemental Materials
A Pilot Study of Visual, Vocal, and Verbal Markers of PTSD
This site hosts the supplemental data, analysis reports, and source code for:
Yang, Y., Jun, D., Welch, B. M., Sylvia, A., Sprunger, J. G., & Girard, J. M. (2026). A pilot study of visual, vocal, and verbal markers of PTSD [Manuscript submitted for publication]. Department of Psychology, University of Kansas.
A preprint of the article is available on Zenodo.
Archived Version
Frozen, citable copies of these materials are archived on Zenodo. The version cited in the article is v1.1.0 (doi:10.5281/zenodo.23266798), which updates the figures archived in v1.0.0; the analyses and results are unchanged.
Supplemental Tables
| Report | Description |
|---|---|
| Sample Demographics | Demographic and clinical characteristics by group, with group comparisons (Table 1) |
| Behavioral Feature Descriptives | Means and standard deviations of all 27 behavioral features by group and task (Table S1) |
Analysis Reports
Each report is a rendered Quarto document containing the complete analysis code and output: model specifications and priors, full parameter tables, the group and interaction effects reported in the article, figures, and posterior predictive checks. The corresponding source files (.qmd) are available in the source folder.
| Report | Description |
|---|---|
| Verbal Feature Analyses | Bayesian mixed models for word count, seven LIWC-22 categories, and LLM-rated sentiment (Table 2, Figure 1) |
| Vocal Feature Analyses | Bayesian mixed models for nine prosodic and voice quality features (Table 3, Figure 2) |
| Visual Feature Analyses | Bayesian mixed models for five facial action units and four head pose and motion features (Table 4, Figure 3) |
Data Availability
The deidentified behavioral features and model covariates analyzed in this work are openly available (CC BY 4.0). Each file has one row per participant and task, and participants are identified only by arbitrary codes (P001–P092) that are shared across files. The codebook describes every column.
The complete v1.1.0 archive (data, code, and reports) can also be downloaded as a single file from Zenodo: Download all (ZIP, 5.9 MB)
These files, together with the analysis source code, are sufficient to reproduce every model, table, and figure in the analysis reports. The video, audio, and transcript data cannot be shared because they contain identifiable protected health information. The other participant-level demographic and clinical measures summarized in Table 1 are not shared to protect participant privacy; the Sample Demographics report shows the code used to produce that table.
Contact
Jeffrey M. Girard, Department of Psychology, University of Kansas — jmgirard@ku.edu