Speaker: Krista Gile, Professor of Mathematics and Statistics, University of Massachusetts Amherst
Description: In 2025, data abound. The day-to-day activities of large parts of the human population leave extensive data footprints. Left out of this data revolution, however, are populations and human experiences that are not represented in standard digital footprints, and especially those that are also beyond the reach of standard survey practices. This includes populations, behaviors, and experiences that are hidden or marginalized in traditional datasets. Examples include populations defined by stigmatized activities that put them at risk for HIV/AIDS including men who have sex with men, people who inject drugs, and female sex workers.
This webinar introduces specialized statistical approaches designed to study such populations. It begins with two methods designed to estimate the population size: the Network Scale-Up Method and Multiple Systems Estimation. It then introduces Respondent-Driven Sampling: a method designed to learn features of the target population (rather than just its size).
The webinar will introduce the basic statistical background of each method, then focus on the hooks: what are the foundational assumptions that allow for inference in these settings of very limited data? We will reflect on the challenges to these assumptions posed by real human populations and reference statistical developments that allow these assumptions to be relaxed or changed. This webinar may be of interest to statisticians interested in an introduction to these 3 methods, interested in studying hard-to-reach human populations, or interested in finding the hook in other challenging data settings.
Date: Friday, November 7, 2025
Time: 2:00pm-3:30pm ET
Format: Virtual/Zoom