Panelynx
Scientific Hiring Method

Hiring decisions backed by science, not gut feeling

Every AI evaluation includes a reasoning chain you can audit. Every score includes a confidence level. Every recommendation is validated against real hiring outcomes. This is not a black box. This is peer-reviewable hiring science.

The 7 Pillars of Scientific Hiring

Each pillar is built into the platform and measured continuously.

Evidence-Based Evaluation

Questions mapped to validated competency frameworks. Expected answers calibrated against real performance data. Difficulty levels validated against candidate pass rates.

Question Bank with expected answers and difficulty tracking

AI Calibration and Accuracy

AI scores compared against expert panelist consensus. Calibration reports show AI vs human agreement rates. Drift detection when AI scoring diverges from panel patterns.

AI Calibration Dashboard in Analytics

Bias Auditing

Statistical analysis of score distributions across candidate demographics. Adverse impact ratio calculation per pipeline stage using the EEOC 4/5 rule.

Bias Audit Tab with compliance indicators

Outcome Validation

Post-hire performance tracking at 90-day, 6-month, and 12-month checkpoints. Correlation analysis: which interview scores predicted actual performance.

Hire Outcomes with auto-reminders

Explainability and Transparency

Every AI evaluation includes a step-by-step reasoning chain. 'Why this score?' audit trail for any AI decision. Full transparency for compliance and trust.

Reasoning chains on every AI evaluation

Confidence Scoring

AI reports confidence levels with every score. Low-confidence evaluations flagged for mandatory human review. Aggregate confidence on interview recommendations.

Color-coded confidence badges with review flags

Continuous Improvement

Questions ranked by predictive validity. Low-effectiveness questions flagged for retirement. AI plan generation prefers high-validity questions.

Question Effectiveness ranking in Question Bank

Black Box AI vs Validated AI

CapabilityTypical AI ToolsPanelynx
AI scores with reasoning chain
Step-by-step evaluation logic
Confidence levels on every score
Statistical bias auditing (4/5 rule)
Post-hire outcome validation
Question effectiveness ranking
AI vs panelist calibration metrics

What We Measure

4/5 Rule

Adverse Impact Compliance

EEOC standard for detecting selection bias per demographic group

r = 0.XX

Predictive Validity

Pearson correlation between interview scores and post-hire performance

90d/6m/12m

Outcome Checkpoints

Post-hire performance tracked at three milestones

The Methodology Engine

Under the hood, every AI-generated plan, question, and evaluation is constrained by a scientific framework. The AI is an execution tool, the science is constant.

Competency Framework

Every job description is mapped to a standardized competency taxonomy combining O*NET knowledge domains and Bartram's Great Eight behavioral competencies. This ensures consistent, research-backed evaluation criteria regardless of how the original JD was written.

30+ technical competencies from O*NET8 behavioral categories from BartramProblem-solving and analytical reasoning

Bloom's Taxonomy Calibration

Question difficulty is calibrated using Bloom's revised taxonomy (1956, 2001). Junior candidates face Remember and Understand questions. Senior candidates face Evaluate and Create. This prevents under-challenging experienced candidates and over-challenging newcomers.

1. Remember2. Understand3. Apply4. Analyze5. Evaluate6. Create

Round-Type Intelligence

Each interview round type has scientific rules about what to assess and how. A Technical Screen tests domain knowledge. A Behavioral round uses STAR questions for past experience. A Culture Fit round avoids technical domains entirely. The AI follows these rules precisely.

Technical Screen: Domain knowledge focus, JOB_KNOWLEDGE questions, Bloom Level 3-5
Behavioral Round: Past experience focus, STAR format, competency-based
Phone Screen: Motivation and basic fit, lighter Bloom levels

BARS Rubrics

Every generated question includes a 5-point Behaviorally Anchored Rating Scale (Smith & Kendall, 1963). Each score level describes specific observable behaviors the candidate says or does, eliminating subjective scoring.

1 - Unsatisfactory2 - Below Average3 - Competent4 - Good5 - Excellent

Automated Bias Detection

Every AI-generated question, rubric, and screening criterion passes through a 6-check bias detection pipeline before reaching the user. Flagged content is surfaced with specific remediation suggestions.

Protected class references (age, gender, race, religion)
Cultural bias (institution-specific, region-specific assumptions)
Socioeconomic bias (asset requirements, relocation demands)
Gender-coded language (Gaucher et al. 2011 validated word lists)
Experience bias (specific employer or exact year requirements)
Disability bias (physical ability assumptions beyond job functions)

Built on Research

Our methodology draws from decades of industrial-organizational psychology research.

The Validity and Utility of Selection Methods in Personnel Psychology

Schmidt, F.L. & Hunter, J.E. (1998)

Structured Interviewing: Raising the Psychometric Properties of the Employment Interview

Campion, M.A., Palmer, D.K., & Campion, J.E. (1997)

re:Work Guide: Structured Interviewing

Google People Operations (2015)

Uniform Guidelines on Employee Selection Procedures (4/5 Rule)

U.S. Equal Employment Opportunity Commission (1978)

Evidence of a New Construct: The Measurement of Gendered Wording in Job Advertisements

Gaucher, D., Friesen, J., & Kay, A.C. (2011)

Taxonomy of Educational Objectives: The Classification of Educational Goals

Bloom, B.S., Engelhart, M.D., et al. (1956)

The Great Eight Competencies: A Criterion-Centric Approach to Validation

Bartram, D. (2005)

O*NET: The Occupational Information Network

U.S. Department of Labor (2023)

The Development and Evaluation of Behaviorally Anchored Rating Scales

Smith, P.C. & Kendall, L.M. (1963)

Consequences of Individual Fit at Work: A Meta-Analysis of Person-Job, Person-Organization, Person-Group, and Person-Supervisor Fit

Kristof-Brown, A.L., Zimmerman, R.D., & Johnson, E.C. (2005)

Companies using structured hiring reduce bad hires by 74%

Ready to fix your interviews?

Join teams who've replaced gut-feeling interviews with structured hiring that finds the right people, every time.

No credit card required. Free forever plan available.