# MARION MORANETZ (541) 507-7549 | Marion.m.moranetz@gmail.com | linkedin.com/in/melmarion github.com/melmarion | Open to Relocation --- Applied behavioral scientist who builds measurement systems for human decision-making. Architect of Persuasion-Max, a 5-layer persuasion prediction engine calibrated on 126K human interactions that quantifies WHY users respond to specific message framings. 80 repositories spanning behavioral measurement, interaction modeling, and native product development. Looking for Quantitative UX Research roles where behavioral science drives product decisions through rigorous measurement. --- ## RESEARCH & MEASUREMENT ### Persuasion-Max — Behavioral Measurement Engine | 2026 *5-layer pipeline that predicts user response to messaging by decomposing persuasive interactions into measurable cognitive and behavioral components. Open-source, calibrated on real human decision data.* - **Designed 16-dimension recipient modeling system** (Big Five personality, Moral Foundations, cognitive style, ideology) that predicts behavioral response variance. Discovered 21.8pp compliance spread across persona types — proving that user response is recipient-dependent, not message-universal. - **Built and validated measurement pipeline on 126K human interactions** across DailyPersuasion (78K dialogues, 35 domains) and HumanChoicePrediction (48K real binary decisions). Discovered that linguistic surface features contribute +10pp AUC over cognitive appraisal alone; binary technique detection adds zero incremental signal. - **Quantified interaction effects between message techniques and user personality** using 400-cell interaction matrix. Moral reframing (Feinberg & Willer 2015) produces +8.4pp lift when matched to recipient's moral foundations — zero lift when mismatched. - **Validated context-dependency of behavioral models**: cross-domain weight transfer degrades AUC by 10-13pp, empirically confirming that user behavior models must be domain-specific, not universal. - **329 automated tests, 302 documented parameters** with full provenance tracking (1% fitted, 61% constrained from literature, 37% uncalibrated). Ablation analysis quantifying each layer's marginal contribution. ### User Behavior Research Across 80 Products | 2025-2026 - **Built 4 iterative versions of social interaction measurement tools** (Charisma Training Game → Read the Room → SWAY → TELLS), each version refining how conversational skill is measured and scored. Final version scores responses across 5 behavioral dimensions with per-scenario calibration. - **Designed behavioral engagement measurement for 15+ games and apps**, implementing variable reward schedules, session architecture (11-minute model), streak mechanics, and DOSE neurochemical framework. Each mechanic's retention impact quantified in design documentation. - **Created 27 reusable behavioral measurement frameworks** (Claude Code skills) codifying engagement scoring, gamification architecture, cognitive flow measurement, and sensory feedback calibration. These function as standardized measurement instruments for product behavioral analysis. ### Content Effectiveness Measurement Platform | 2026 - **Built closed-loop measurement system** correlating predicted technique effectiveness against observed engagement metrics across Reddit, LinkedIn, and Substack. Auto-updates effectiveness ratings after 20+ observations per technique. - **Engineered 12 community linguistic profiles** measuring reading difficulty, emotionality, and pronoun ratios. Content deviating >1.5 sigma from community norms flagged automatically — quantifying "fit" rather than relying on subjective review. --- ## PRODUCT DEVELOPMENT - **34 iOS applications (Swift/SwiftUI)** including persuasion measurement trainers, narrative experiences, creature/ecology simulations, and ambient wellness tools. Each built with behavioral measurement as core architecture. - **39 browser-based projects (React/Canvas)** including influence detection tools, generative experiences, and a 14-game casino platform (ARCANE) with provably fair outcome verification. --- ## TECHNICAL PROFICIENCY **Measurement & Analysis:** A/B testing, bootstrap confidence intervals, ML evaluation (AUC/ROC, cross-validation), interaction term discovery, ablation analysis, NLP feature engineering, survey design **Behavioral Science:** Cognitive appraisal theory (Smith & Ellsworth), Moral Foundations Theory, Elaboration Likelihood Model, Big Five personality modeling, DOSE neurochemical framework, persuasion technique taxonomies (Zeng 40-technique / SemEval) **Languages & Frameworks:** Python, Swift/SwiftUI, JavaScript/React, scikit-learn, FastAPI, Canvas API, SQL **ML/AI:** LLM fine-tuning (Ollama/7B), Claude API, logistic regression pipelines, NLP, feature selection, prompt engineering **Tools:** Git (80 repos), Xcode, VS Code, Claude Code, MCP protocol --- ## EDUCATION **Southern Oregon University** — Ashland, OR | 2025 B.S. Innovation & Leadership (GPA: 3.76) | Minor: Rhetoric & Reason - Coursework: Persuasion & Negotiation (A), Data-Driven Decision Making, AI/ML Applications **Southwestern Oregon Community College** — Coos Bay, OR | 2023 A.S. General Studies (GPA: 3.8, top 5%) | Completed 2-year program in 1.1 years - Academic Mentor to 50+ students. Student Government Senator: reversed 2-year attendance decline with behavioral event design. --- ## KEY METRICS - 21.8pp persona compliance spread in technique x personality interaction matrix - 10-13pp AUC degradation on cross-domain transfer, validating domain-specific modeling - 126,288 human interaction records processed with full parameter provenance - 414 automated tests across 2 primary codebases, all passing - 302 model parameters with documented provenance (literature-constrained vs fitted)