Assistant Professor of Computational Social Science
brosche A VS Code chat extension that gives students hints, not answers. It fetches the reference solution from a private instructor repo, hides it from the student, and constrains the model to ask diagnostic questions and escalate hints across turns rather than reproducing the solution.
Grading student work is hard, and grading the reasoning behind it is even harder. A correct answer rarely shows real understanding, while written explanations are slow to review at scale. I’ve been exploring whether a short, structured conversation with an LLM — asking students to explain their reasoning — can complement submitted work and provide instructors with a useful second signal. This tutorial walks through a prototype open-source app that lets students upload work, chat briefly with an AI, and generate advisory per-dimension grades for instructors to review and override.
Coalition research increasingly emphasizes party-level explanations of coalition outcomes. However, this work does not account for the complex multilevel structure between parties and governments: many parties participate in multiple governments and governments often comprise multiple parties. In this paper, I show that this crisscrossing structure creates dependencies among observations both across and within governments. If … Read more
Explaining how macro-level outcomes emerge from their constituting parts at the micro-level is a complex undertaking. In empirical research, however, statistical methods that feature trivial aggregation functions dominate because methods to study more complex aggregation processes remain underdeveloped. In this thesis, I contribute to the development of empirical-statistical methods for the study of micro-macro links. … Read more
Adolescent friendship networks exhibit limited interaction across socioeconomic and racial lines. Using Add Health data and a novel exponential random graph model, this study examines socioeconomic segregation in high school friendships and its relationship to racial segregation. Results show that networks are segregated less by socioeconomic status (SES) than by race, yet low-SES students are … Read more
The world and data about the world are becoming increasingly complex. Examples of complex data structures include network data that represent connections among individuals (e.g., friends on social media platforms), spatial data that represents geolocations (e.g., smartphone location data), data collected at multiple levels (e.g., employees in organizations), and text data (e.g., interviews, online comments). These … Read more
Most courses on social network analysis (SNA) focus on descriptive SNA, such as measuring the density of a network, identifying subgroups within a network, or examining the centrality of actors within a network. Inferential SNA, which focuses on explaining the formation of networks and behaviors and beliefs of actors embedded in networks, by contrast, is … Read more
Many kinds of data have a multilevel, hierarchical, nested, or clustered structure. Animal and human studies of inheritance, for instance, deal with a natural multilevel structure where offspring are grouped within families. Offspring from the same parents tend to be more alike than individuals at large. Clinical trials and other data collections also create data … Read more
Despite much research effort into response enhancing methods, trend studies over the years showed that response rates are declining. Differences in nonresponse trends over countries can only partially be explained by differences in survey design and field methods between countries. General attitudes towards surveys and survey climate are often named as important theoretical concepts for … Read more