Overview
Benjamin Rosche is a computational social scientist with a background in sociology, economics, statistics, and computer science. Ben leverages and develops quantitative methods to analyze family and network dynamics as drivers of social inequality. Methodologically, he specializes in the modeling of complex dependencies in observational data (esp. multilevel, spatial, and network embeddedness). Next to his studies in sociology, Ben pursues a minor in the Department of Computer Science with a focus on machine learning and natural language processing and works for the OECD International Programme for Action on Climate (IPAC) as a data science consultant. His research is funded by the National Science Foundation.
Subfields
Social Networks, Social Inequality, Family Demography, Quantitative Methods (esp. inference with multilevel, spatial, network, and textual data)
Dissertation Committee
Chair: Michael Macy (Cornell Sociology), Filiz Garip (Princeton Sociology), Felix Elwert (UW-Madison Sociology), Eleonora Patacchini (Cornell Economics), Lillian Lee (Cornell Computer Science)
Research Agenda
Network dynamics as drivers of social inequality
This project addresses three related questions: Do disadvantaged youth befriend peers from higher socioeconomic backgrounds? Do they benefit from such ties that cross socioeconomic boundaries? And do the benefits depend on the structure of their friendship networks?
To answer these questions, Ben draws on modern network analysis methods and data from the Add Health panel, which includes direct measures of peer attributes, longitudinal friendship network data, and tracks adolescents far enough into adulthood to examine their long-term socioeconomic attainment. The project has received a three-year NSF award ($240,000; PI: Benjamin Rosche) in collaboration with Michael Macy (Cornell Information Science), Eleonora Patacchini (Cornell Economics), and David Grusky (Stanford Sociology).
As part of this project, Ben collaborated with Weihua An (Emory Sociology) on a review paper on causal network analysis methods, which was published in the Annual Review of Sociology.
The first paper in this project (job market paper), titled “Socioeconomic segregation in adolescent friendship networks: Determinants of ties within and between socioeconomic boundaries.”, uses recent advances in decomposition methods for exponential random graph models to measure socioeconomic segregation in high school friendship networks more directly and decompose its determinants more comprehensively than prior research. The results show that adolescent friendship networks are bifurcated in that students from low socioeconomic backgrounds are separated from other students. The analysis of determinants reveals that this separation is not driven by selection processes into courses or extracurricular activities, but instead by racial and socioeconomic homophily within those settings. Consequently, while de-tracking promotes educational equality, it is not an effective tool to integrate friendship networks. Accordingly, interventions to reduce socioeconomic segregation in friendship networks should focus on attenuating students’ homophilous tendencies. The analysis of determinants further indicates that network mechanisms, such as triadic closure, have an ambivalent impact on segregation in that they foster both ties within and between socioeconomic boundaries. This finding challenges the widely-held belief that network mechanisms amplify segregation in homophilous networks (e.g., Kossinets and Watts 2009; Asikainen et al. 2022).
- Robin M. Williams Jr. Best Paper Award, Department of Sociology, Cornell University, 2023
Family dynamics as drivers of social inequality
There is an extensive body of literature on the motherhood penalty. By contrast, much less is known about how the motherhood effect has shaped women’s earnings in ways that contribute to aggregate inequality. Using U.S. Census data, this project examines how the changing effect of motherhood on women’s earnings between 1980 and 2020 has impacted women’s earnings inequality.
The first paper of this project, titled “Treatment effects on within-group and between-group inequality. A causal variance decomposition approach.”, introduces a new approach to examining how treatment variables impact within-group, between-group, and total inequality. The approach is applied to examine the changing effect of motherhood on women’s earnings and its consequences for women’s earnings inequality between 1980 and 2020. The results show that changes in the motherhood effect since the 80s have reduced inequality in women’s earnings. The decomposition indicates that motherhood primarily reduces inequality within economic strata. The motherhood effect on inequality between economic strata is small relative to the within-group effect. This result highlights the risk of drawing misleading conclusions when inequality researchers base their analyses solely on mean differences between groups.
Other Research
A multilevel model to study aggregation processes
The paper, titled “A multilevel model for coalition governments: Uncovering dependencies within and across governments due to parties.”, advances a multilevel model to study aggregation processes. Most multilevel analyses examine how lower-level units (e.g., people) are affected by their embedding in contextual/aggregate units at a higher level (macro-to-micro link). This paper develops a generalization of the multiple membership multilevel model (MMMM) to conceptually reverse this setup. The model allows studying how the effects of units at lower levels aggregate to a higher level (micro-to-macro link). The proposed model—akin to a spatial autoregressive model with endogenized weights—offers an empirical approach to the study of aggregation problems. Previous studies examining micro-to-macro links either aggregated or disaggregated the data. These approaches obstruct the inherent aggregation problem, cannot separate micro-level and macro-level variance, and ignore dependencies among observations, thus inducing excessive Type-I error. The MMMM overcomes these problems by explicitly modeling the aggregation from the micro to the macro level by including an aggregation function in the regression model.
The model’s utility is demonstrated with an application to coalition government survival as predicted by parties’ financial dependencies. The results show that the more parties’ financial resources comprise member contributions, the higher the termination hazard of governments including those parties. The aggregation analysis indicates that the effect of parties on the termination hazard is proportional to their seat share in parliament.
- Robert B. McGinnis Best Methods Paper Award, Department of Sociology, Cornell University, 2020
Publications
Peer-Reviewed Publications
An, W., Beauvile, R., & Rosche, B. (2022). Causal Network Analysis. Annual Review of Sociology, 48.
* The authors are listed alphabetically and contributed equally to the article
Work in progress
Rosche, B. (2023). Socioeconomic segregation in adolescent friendship networks. Determinants of ties within and between socioeconomic boundaries.
- Robin M. Williams Jr. Best Paper Award, Department of Sociology, Cornell University, 2023
- In preparation for submission.
- In preparation for submission.
- Robert B. McGinnis Best Methods Paper Award, Department of Sociology, Cornell University, 2020
- In preparation for submission.