Research Portfolio
This is a portfolio of research questions I’ve answered and the ones I’m still working on. The work clusters into three interconnected strands: education policy and data science, writing assessment and AI, and cultural analytics.
🗂️ Mission Log
Last 10 notes on 🗂️ Mission Log
Latest posts
Kanban
not started doing What would an open standard for tutoring data look like What’s so high about tutoring anyway—dosage, quality, impact? Do cosine-similarity metrics correctly ...
Presentations
Conference papers, keynote talks, and invited presentations from Jason Godfrey's curriculum vitae.
How usable is the federal ESSER expenditure dataset for research?
A public correction notice for the AERA 2026 conference analysis.
Are original films more unequal than franchise films?
Read it in Reel Metrics → Short answer: Yes — and the gap tripled in 25 years. Among U.S. wide releases (500+ peak theaters), the franchise Gini rose from 0.44 (2000–07) to 0.57 (2021–25).
Will new grants without measurement systems work?
Read it in The 74 Million → Short answer: No — not unless we build the measurement infrastructure first. New federal education grants are about to be awarded to states that have no consistent way to track whether the money produces results.
Why is scaling high-dosage tutoring a measurement problem?
Read it on Accelerate → Short answer: The instructional case for high-dosage tutoring is settled; the bottleneck is visibility. Two AERA 2026 presentations — one on what five states have learned from two years of implementation, the other on the quality of the federal ESSER expenditure record — converge on the same point: without interoperable, trustworthy data, neither states nor researchers can…
How did Disney come to dominate the box office?
Read it in Reel Metrics → Short answer: Disney’s acquisitions of Pixar (2006), Marvel (2009), Lucasfilm (2012), and Fox (2019) roughly doubled its domestic share from ~14% (2000-07) ...
Why are movies getting longer?
Read it in Reel Metrics → Short answer: Four follow-ups to Stephen Follows’s runtime analysis using the CPRF Movie Database. Longer films have better legs but only up to about 135 minutes.
Does an Oscar nomination boost box office revenue?
Read it in Reel Metrics → Short answer: Nominations redirect revenue from non-nominees to nominees but do not expand the total market. A regression discontinuity design across 27 ceremony years (2000-2026) finds a market share shift of +4.1 percentage points (p=0.009) with no evidence of total market growth.