Data Practices and Privacy
Data Collection
PSI collects data through five assessment modules:
| Module | Data Type | What's Captured |
|---|---|---|
| A | Text + Timing | Written responses to video scenarios, response latency |
| B | Selection + Timing | Microexpression recognition choices, response speed |
| C | Audio | Voice recordings (30-second voicemails) |
| D | Text | Written email responses |
| E | Likert Responses | Self-perception ratings |
Data Processing and Retention
Audio Processing (Module C only)
- Voice recordings are processed using multimodal AI analysis that evaluates vocal qualities holistically (warmth, pace, composure, energy)
- Raw audio files are deleted immediately after scoring completes—within the same workflow execution
- What persists: dimensional scores and categorical vocal analysis, not acoustic waveforms or transcripts
- No biometric templates are stored for identification purposes
Text and Selection Data
- Written responses are analyzed for linguistic patterns, then stored in anonymized form
- Responses are not linked to external identifiers beyond assessment session
Encryption
- All data encrypted in transit (TLS 1.3)
- All data encrypted at rest (AES-256)
What PSI Does NOT Do:
• No facial recognition or storage — Module B shows faces to participants; participants' faces are not captured
• No continuous biometric monitoring — Audio analysis occurs only during 30-second recording tasks
• No cross-session tracking — Each assessment is independent
• No selling of data — Assessment data is not sold or shared with third parties for marketing
Assessment Variants
PSI-Quick (Abbreviated Assessment)
| Feature | Full PSI | PSI-Quick |
|---|---|---|
| Duration | ~20 minutes | ~8 minutes |
| Modules | A, B, C, D, E | A (3 scenarios), D (1 email) |
| Dimensions | All 7 | All 7 (via Module A) |
| Confidence | Calibrated | Low (limited data points) |
| Use Cases | Development planning, coaching | Screening, quick pulse |
The Science Behind PSI-7
Each PSI dimension maps to established psychological research while focusing on observable, trainable capabilities.
1 Situational Radar
The ability to perceive emotional context, read nonverbal cues, and sense unspoken dynamics.
Research Foundation
- Emotional Perception (Mayer & Salovey, 1997): Accurate perception of emotions is the foundational layer of emotional competence
- Facial Action Coding System (Ekman & Friesen, 1978; Ekman, 2003): Standardized emotional expression categories as trainable recognition targets
- Social Cognition (Adolphs, 2009): Social perception operates through specific brain networks; targeted practice improves accuracy
AI Scoring Signals
| Signal | Source | Measurement |
|---|---|---|
| Microexpression accuracy | Module B | Correct identification percentage |
| Confusion patterns | Module B | Which emotions misidentified for which |
| Response calibration | Module B | Speed-accuracy tradeoff patterns |
| Contextual awareness | Module A | Recognition of emotional context in responses |
2 Emotional Regulation
The ability to manage emotional responses and maintain composure under pressure.
Research Foundation
- Emotion Regulation Theory (Gross, 1998): Five families of regulation strategies deployable based on context
- Stress Response Research (Sapolsky, 2004): Acoustic patterns provide objective stress indicators
- Resilience and Recovery (Fredrickson et al., 2003): Regulation capability is trainable through deliberate practice
AI Scoring Signals
| Signal | Source | Measurement |
|---|---|---|
| Vocal composure | Module C | Steadiness and stability under pressure |
| Vocal warmth | Module C | Tonal patterns conveying empathy and connection |
| Speech pace | Module C | Natural rhythm versus rushed or hesitant delivery |
| Response composure | Module A | Calm versus reactive text patterns |
3 Outcome Ownership
The projection of competence, accountability, and agency.
Research Foundation
- Locus of Control (Rotter, 1966): Internal locus correlates with proactive behavior and achievement
- Competence Projection (Anderson & Kilduff, 2009): Specific commitments and active voice shape competence perceptions
- Action Orientation (Kuhl, 1994): Distinguishes those who move toward goals from those focused on obstacles
AI Scoring Signals
| Signal | Source | Measurement |
|---|---|---|
| Agency language | Modules A, D | "I will" vs. passive constructions |
| Commitment specificity | Modules A, D | Concrete actions vs. vague intentions |
| Active voice usage | Modules A, D | Subject-verb-object clarity |
| Hedging absence | Modules A, D | Minimal qualifiers and deferrals |
Score Architecture: The Living Profile
PSI implements a four-layer score model that distinguishes between different types of measurement:
| Layer | Name | What It Measures | When It Changes |
|---|---|---|---|
| 1 | Baseline | First full PSI assessment | Fixed after initial assessment |
| 2 | Self-Perception | Module E self-assessment | Each new self-assessment |
| 3 | Performance | Modules A-D behavioral scores | Each new full assessment |
| 4 | Current | Living score (baseline + training) | After practice sessions |
Training Weight Ramping
- Sessions 1-2: 100% baseline, 0% training data
- Sessions 3+: Training weight increases gradually
- Maximum: 80% training weight after sustained practice
Awareness Gap Calculation
- Gap = Self-Perception (Layer 2) − Performance (Layer 3)
- Gap threshold: ±15 points triggers blind spot flagging
- Positive gap (self > performance): Overconfidence
- Negative gap (self < performance): Underconfidence
Competitive Comparison
| Feature | PSI | DISC | MBTI | EQ-i 2.0 | Big Five |
|---|---|---|---|---|---|
| Data Collection | Multimodal | Self-report | Self-report | Self-report | Self-report |
| What's Measured | Demonstrated capability | Declared preference | Cognitive preferences | Self-reported EI | Personality traits |
| Blind Spot Detection | Yes | No | No | Limited (360) | No |
| Predictive Validity | 0.26-0.42* | Limited evidence | r ≈ 0.10-0.20 | r ≈ 0.20-0.30 | r ≈ 0.19-0.31 |
*PSI uses behavioral assessment methodology. Validity estimates based on Sackett et al. (2022) meta-analysis of behavioral methods generally; PSI-specific validation ongoing.
Validation Approach
Construct Validity
PSI dimensions map to established constructs from emotional intelligence, communication competence, and leadership effectiveness research. Factor analysis during development confirmed that the seven dimensions capture distinct but related capabilities.
Criterion Validity
Ongoing research examines correlations between PSI scores and:
- Performance ratings (manager and peer)
- 360-degree feedback scores
- Engagement and retention metrics
- Leadership advancement
Face Validity
Participant feedback consistently confirms that PSI dimensions represent meaningful, recognizable interpersonal capabilities.
Key References
Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection. Journal of Applied Psychology.
Mayer, J. D., & Salovey, P. (1997). What is emotional intelligence? Emotional Development and Emotional Intelligence.
Ekman, P. (2003). Emotions Revealed. Times Books.
Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly.
Fiske, S. T., Cuddy, A. J. C., & Glick, P. (2007). Universal dimensions of social cognition: Warmth and competence. Trends in Cognitive Sciences.
Gross, J. J. (1998). The emerging field of emotion regulation: An integrative review. Review of General Psychology.
Kruger, J., & Dunning, D. (1999). Unskilled and unaware of it. Journal of Personality and Social Psychology.
Lacerenza, C. N., et al. (2017). Leadership training design, delivery, and implementation: A meta-analysis. Journal of Applied Psychology.