Research

Scientific productivity as a stochastic process

My current research examines how scientific productivity changes over the course of a career. I develop scholar-agnostic stochastic models that aim to reproduce observed features of publication trajectories without assuming stable, inherent differences in ability between researchers.

This work focuses on:

• self-exciting and history-dependent processes
• extensive and intensive margins of productivity
• rank persistence and rank mixing
• heavy-tailed year-to-year changes
• temporary inactivity, restart, and dropout
• generative model comparison

Complex systems

More broadly, I am interested in how persistent individual trajectories emerge from interactions among stochastic dynamics, accumulated history, social structure, and institutional opportunity.

Current project

My current project extends earlier random-walk models of scientific productivity by comparing autoregressive, hurdle, and self-exciting models against empirical career trajectories.

The project evaluates whether parsimonious dynamic models can jointly reproduce:

• mean and variance trajectories
• productivity increments
• cumulative productivity distributions
• career-stage transitions
• rank persistence
• periods of inactivity and return

Methods

My work uses tools from:

• stochastic processes
• time-series analysis
• computational social science
• statistical inference
• simulation-based model evaluation
• complex-systems science

Code

Code and computational materials are available through my GitHub profile.