Blog

Notes on cricket, handball, and statistical methodology — mostly work in progress, thinking out loud.

Notes on statistical methodology — how to design studies, measure reliably, and trust the numbers.

Does the room change the speech?

elections
measurement
LLM
methodology

What happens when you score every campaign speech for populist rhetoric and then ask where each one was delivered. A measurement study of audience design in U.S. presidential campaigns.

Aug 27, 2026
3 min

How many innings before a batting average means anything?

cricket
G-theory
measurement

Cricket’s favorite number is also its most misread. A measurement-theory look at why small-sample averages rank noise — and how many innings it takes to fix that.

Aug 19, 2026
3 min

Testing one thing at a time is costing you

methodology
design of experiments

One-factor-at-a-time experimentation feels careful and rigorous. It’s actually the expensive way to learn less — here’s the arithmetic.

Aug 19, 2026
3 min

How many participants do you actually need?

methodology
study design
grants

The most common question a statistician gets — and why the honest answer starts with a different question: what’s the smallest effect you’d hate to miss?

Aug 19, 2026
3 min

Your leaderboard gap is probably noise

AI
methodology
measurement

Two models, two percentage points apart, thousands of votes — and still not enough evidence to call a winner. How to read a ranking like a statistician.

Jul 22, 2026
3 min

Building this site: Quarto, GitHub Pages, and a win-probability worm

meta
R
cricket

Why I picked a plain-text, version-controlled site over a CMS — and how the cricket forecaster on the home page connects to a real model.

Jun 13, 2026
1 min

How many observations before a statistic means anything?

G-theory
methodology
sports

Generalizability theory’s decision study — the tool for knowing when a performance metric is signal and when it’s still noise.

Jun 8, 2026
2 min
No matching items