Matthew Nelson, PhD

Reputation research for Amazon's Reputation Marketing & Insights team, covering Amazon Retail and AWS across global markets.

• Conduct global survey research and advanced analytics across Amazon Retail and AWS, measuring brand reputation, executive favorability, competitive positioning, and perceptions of Amazon as an AI leader
• Develop methodology across countries: sampling plans, quota frameworks, questionnaire design, fieldwork specifications, and weighting to census benchmarks to ensure comparable results across markets and over time
• Model reputation outcomes using logistic regression, Shapley decomposition, and specification curve analysis to identify drivers of brand perception
• Build research infrastructure: automated Python pipelines and dashboards that turn raw fieldwork into crosstabs and tracking views, plus an internal report-writing tool built on the Claude API with custom steering files for drafting briefs and number checking
• Interview research and engineering candidates; mentor and evaluate a research scientist intern

Quantitative research for Fortune 50 companies navigating reputation risk, competitive pressure, and crisis communications. Isolate what drives a reputation, then test which messages move it.

• Designed survey instruments and statistical models for Fortune 50 clients across energy, pharmaceuticals, consumer products, manufacturing, and technology
• Modeled reputation drivers with logistic regression and built a driver analysis mapping each attribute by its predictive strength and association with the brand
• Ran multi-wave message and creative testing across distinct stakeholder audiences, combining max-diff with dial-tested focus groups
• Produced research briefs and executive presentations translating quantitative findings into strategic recommendations
• Led firm-wide training on quantitative methods and AI tooling, and helped develop the AI policy governing data safety and researcher accountability

Research on electoral institutions, redistricting, and voting rights, published in peer-reviewed journals and cited before the U.S. Supreme Court.

• Published in PS: Political Science & Politics; cited in Supreme Court amicus brief, Moore v. Harper
• Applied regression analysis, ecological inference, and racially polarized voting methods to help guide redistricting in Riverside County
• Research covered by the Washington Post, FiveThirtyEight, and CalMatters