Neil Fasching

Computational Social Scientist

Quantitative researcher with industry experience at Meta and a PhD from the University of Pennsylvania. I specialize in survey design, experimental methods, and causal inference. My research spans public opinion, political communication, and AI bias and fairness, with publications in Science Advances, ACL, and the British Journal of Political Science.

Download CV
See my research → Google Scholar →
Neil Fasching
01

Experience

Quantitative UX Researcher
Meta — Central Growth
Mar – May 2026
  • Updated the “New Connections” off-platform survey, then used Claude Code and OpenAI Codex to modernize the analysis, building automated longitudinal workflows across previously siloed survey waves
  • Designed a quantitative on-platform Friending Intent survey, with results to directly inform the Friending team’s decision to update Facebook’s People You May Know (PYMK) algorithm
  • Smoke-tested the uxr-toolkit research infrastructure (Claude Code slash command), helping to design quantitative research workflows that could drive efficiency gains across Meta’s UXR Org
  • Helped build and deploy multiple internal AI agents to accelerate research workflows, including a personal knowledge agent (“Second Brain”), a shared team agent (“Team Brain”), and a custom MyClaw instance
Computational Research Fellow
University of Pennsylvania
Sep 2021 – Mar 2026
  • Designed and executed large-scale quantitative studies including surveys, behavioral experiments, and NLP-based analysis to understand how users interact with media, AI systems, and digital platforms
  • Built AI-powered measurement pipelines analyzing over 28,000 podcast episodes and 46.7 million social media posts, delivering findings to interdisciplinary research teams and public-facing stakeholders
  • Co-Instructor for PhD-level Modern Data Mining at Wharton (2022–2026), covering LLMs, neural networks, and ensemble methods
  • Shipped a public-facing research product (mediabiasdetector.seas.upenn.edu) for external audiences
02

Selected Research

2025
Association for Computational Linguistics
Model-Dependent Moderation: Inconsistencies in Hate Speech Detection Across LLM-based Systems
Content moderation systems powered by large language models (LLMs) are increasingly deployed to detect hate speech; however, no systematic comparison exists between different systems. If different systems produce different outcomes for the same content, it undermines consistency and predictability, leading to moderation decisions that appear...
Read paper → Details
2024
Science Advances
Persistent polarization: The unexpected durability of political animosity around US elections
The scholarly literature suggests that, as elections approach, political tensions intensify, and, as they pass, tensions return to pre-election levels. Using a massive new dataset of 66,000 interviews (cross-sectional and panel), we find that animosities are durable and consistent over the course of the 2022...
Read paper → Details
2024
Political Psychology
A critical evaluation and research agenda for the study of psychological dispositions and political attitudes
Political psychologists often examine the influence of psychological dispositions on political attitudes. Central to this field is the ideological asymmetry hypothesis (IAH), which asserts significant psychological differences be- tween conservatives and liberals. According to the IAH, conservatives tend to exhibit greater resistance to change, a...
Read paper → Details
View all research →
03

Methods & Skills

Research Methods

Survey design & fielding · Experimental & quasi-experimental design · A/B testing · Causal inference (ITS, DiD, RDD) · Behavioral data analysis · Mixed methods

Statistical Analysis

Regression (linear, logistic, multilevel) · SEM & factor analysis · Time-series analysis · Bayesian methods · Psychometrics

Programming

Python (Expert) · R (Expert) · SQL (Expert) · JavaScript (Proficient) · PySpark · Pandas · dplyr/tidyverse

AI & Platforms

Claude Code · OpenAI Codex · Agent orchestration · AWS · Microsoft Azure · Git/GitHub · Posit Workbench