Justin Weltz. (image: courtesy Justin Weltz)

Conventional approaches to the study of inequality focus on income and wealth, but an individual’s position within their economic networks is another important form of inequality. Efforts to understand these economic networks rely heavily on conventional surveying techniques, which often fall short when applied to populations defined by a stigmatized behavior or status.

Respondent-driven sampling, which uses referrals, is one sampling technique used to study hard-to-reach populations. For example, to learn about HIV rates among intravenous drug users, researchers might start with a small group of people who already participate in a needle exchange program. Then, the researchers offer the original participants incentives to recruit others into the survey. This process continues — old participants recruiting new participants — until a target sample size or budget is reached. 

“The problem,” says EPE and Applied Complexity Fellow Justin Weltz,“is that the people you are associated with are, by and large, similar to you in some sense. It’s far from a random sample.”

Weltz draws on statistics, reinforcement learning, and network science to develop new methods for gathering quality data about understudied groups. While completing a Ph.D. in statistical science at Duke University, he built tools that he envisions might one day be applied to measure the impact of interventions on infectious diseases and marketing promotions, or track rates of colorectal cancer among people without health insurance.

During his fellowship at SFI, Weltz will continue developing new techniques for studying complex social networks. He will focus on issues of policymaking and wealth inequality, working closely with SFI External Professor Eleanor Power (London School of Economics) on the Economic Networks and the Dynamics Of Wealth (Inequality) — ENDOW — program. “The idea, broadly, is to understand how social networks relate to wealth inequality,” says Weltz. “I am interested in constructing these networks from sample data so that this relationship can be explored.” Weltz arrives in September.