I study how customers, organizations, and markets respond when new technologies create unfamiliar choices. My work asks what people value, how firms position and adopt products, how organizational structures shape innovation, and how policy influences what gets built and scaled.
Across projects, I combine randomized surveys, product-level and longitudinal data, web scraping, interviews, and quasi-experimental analysis. I use these methods not only to explain what happened, but to build measures, test competing explanations, and translate evidence into product, market, commercialization, and policy decisions.
Together, these projects examine technology adoption at three connected levels: customer and product choices; organizational adoption and innovation; and the markets, ecosystems, and policies that shape scale.
How does a new technology become a functioning market? Mapping 2,130 drone companies across Chinese cities, this project follows the emergence of China’s civilian drone industry from early technical experimentation through commercialization and regulation. It shows how the central bottleneck shifted over time, requiring different forms of ecosystem and institutional support at each stage.
Methods: Ecosystem mapping, firm and city-level data, value-chain analysis, policy analysis, and industry interviews.
I expected clearer rules to set drone design free. Instead, tracking 335 agricultural drone models launched between 2000 and 2021, I found that new designs converged around the FAA's 55-pound standard: the rule itself became a design magnet. The exceptions were local. Firms near industry consortia followed the standard even more closely, while firms near farmer-facing knowledge networks kept experimenting. Regulation sets the standard. Unconventional designs survive in the places where builders and users learn together.
Methods: Longitudinal product data, regulatory filings, archival research, web archives, and quantitative analysis.
Why do robots that work in controlled settings struggle on real job sites? Drawing on 30+ interviews, field engagement, and a mapped landscape of 300+ construction robotics companies, this project examines how workflow fit, trust, liability, and buyer readiness shape adoption. The research shows that technical capability is rarely enough: successful firms begin with narrow use cases, build credibility, and redesign the surrounding system as they scale.
Methods: Interviews, ecosystem mapping, field research, and facilitated discovery.
Should an emerging technology be positioned around what customers can do with it—or around its fit with new rules? I built a longitudinal panel of 400+ consumer-drone products on Amazon to examine how market-use and compliance messaging related to marketplace traction before and after a major regulatory shift. The analysis suggests that product messaging worked best when it matched the market’s stage of regulatory uncertainty.
Methods: Product-level marketplace data, longitudinal analysis, text-based measurement, count models, and regulatory analysis.
Interdisciplinary work is often treated as inherently beneficial, but its outcomes depend on how roles, incentives, and evaluation systems are designed. I helped build a mixed-method study combining a longitudinal faculty panel with interviews to examine how organizational structure and career context shape research and innovation pathways. The evidence suggests that interdisciplinarity is not a single organizational condition: cross-boundary roles need aligned incentives, evaluation systems, and coordination support.
Methods: Multi-source data integration, panel and outcome models, interview-guide design, faculty interviews, and mixed-method triangulation.
How do governments invest in emerging technologies when no one knows which industries or interventions will succeed? This project transforms 6,800 policy reports into a dataset of 1,659 state-led experiments, revealing four distinct approaches to innovation policy. The findings show that how states distribute their bets across technologies and partners matters more than how much they experiment, and that cross-sector partnerships can make experimentation more resilient during economic shocks.
Methods: Web scraping, natural-language processing, policy coding, portfolio analysis, economic indicators, and interviews with state officials.
Additional Methods Experience: Earlier research includes a randomized conjoint study of consumer preferences and willingness to pay for an antimicrobial product, integrating survey evidence with cost and sensitivity models to inform feature, pricing, and commercialization decisions.