SODAS Data Discussion #4 (Spring 2024)
Copenhagen Center for Social Data Science (SODAS) aspirers to be a resource for all students and researchers at the Faculty of Social Sciences. We therefore invite researchers across the faculty to present ongoing research projects, project applications or just a loose idea that relates to the subject of social data science.
Two researchers will present their work. The rules are simple: Short research presentations of ten minutes are followed by twenty minutes of debate. No papers will be circulated beforehand, and the presentations cannot be longer than five slides.
Discussion 1
Presenter: Vincent Gadegaard.
Title: Challenges of Heterogeneity in Social Media Behavior Data.
Abstract:
In this data-discussion, I share the tentative approach, data, and results of a work-in-progress paper on political participation on public political facebook posts. Following this introduction to my paper, I discuss my general data challenges, and the troubles using the aggregation of individual level social media behavioral data in answering theoretically driven question. Here I touch upon affordances of social media usage and interpretation of behavior, and open up discussion on how to quantify more qualitatively driven questions in the data.
Discussion 2
Presenter: Jeppe Johansen.
Title: Peer Effects in High School Application.
Abstract:
This paper investigates peer effects in high school applications within the context of the Danish education system. It considers all Danish 9th graders from the period 2015 to 2019, addressing the binary choice problem of whether to apply for high school. This encompasses a total of approximately 300,000 students. The study builds on recent advancements in the econometric literature that exploit insights from game theory, particularly the notion that peer effect games are supermodular and allow for fixed point iteration under an equilibrium selection rule (Boucher et al. 2022; Graham and Pelican 2023). Additionally, this paper reframes the discrete choice problem as a perturbed utility model, which facilitates considerably easier estimation than the method proposed by Graham and Pelican 2023. Specifically, the estimation process iteratively performs a logistic regression and then proceeds to a solution step, iterating to a fixed point in the supermodular game. The estimation method is validated through a Monte Carlo simulation demonstrating that a simple approach of plugging in peer averages will yield biased results. The parameter estimates from the real data align with these findings, and the estimation approach suggested in this paper indicates considerably smaller peer effects than what would be suggested by merely plugging in peer averages. Finally, the structural parameters are used for policy simulation of implementing a GPA cutoff of 7. The results indicate that this policy would lead to a substantial decrease in high school applicants, with peer effects having a significant impact on the reduced application numbers.