Education
Teaching
Numbers do not speak for themselves. These are the tools I use to get judges, funders, and colleagues arguing with the data instead of nodding at it.
More in Telling Your Court's Data Story, Trends in State Courts, 2025. Presented at the National Association for Court Management.
The Hook
I hand a room a number worth arguing about, then let the argument do the work. It usually starts with a rap score and ends with people questioning their own metrics.
The six ideas I come back to most often:
Construct validity Sampling and bias Multidimensional measurement Uncertainty communication Visualization design Narrative framing
Contrivances
Every one below is live and self-contained. Open it and try it: each is built to start an argument, not settle one.
The Hip-Hop Periodic Table
A dataset of 187 hip-hop acts, each scored on five parts of lyricism. It opens with a fast gut rating of a favorite artist, then turns to the real question: what did you just measure? People end up debating construct validity using their own performance measures.
Explore the periodic table →
Companion materials: analytics dashboard · data dictionary
The Barnum Test
A short exercise that builds a personalized profile, asks how accurate it feels, then reveals that everyone got the same profile and the ratings were never used at all. The point is not that people were fooled. It is that a measure everyone scores high on has not measured anyone. Works as a cold open or on its own.
Take the Barnum Test →
How Old Is Old?
Set a slider to the age at which someone becomes "old," "a senior," "too old to drive," or "too old to start over." The answers show how little we agree, and how the law pins hard numbers (18, 50, 65) onto a boundary nature never drew. A study in turning a fuzzy idea into an exact number.
Try How Old Is Old? →
Trust Your Gut?
Review realistic case files, predict who returns to court, then see your accuracy scored against a simple two-variable rule that beats nearly everyone. A lesson in why expert intuition keeps losing to a checklist.
Trust your gut? →
Which Evaluation Should We Do?
Say what you have and what you want to know, and every evaluation approach lands in one of three piles: what fits, what is the wrong tool and why, and what is close, with the gap named and the work to close it spelled out.
Open the evaluation picker →
Which Deliverable Fits?
One set of findings can become a two-page brief, a formal report, a dashboard, or a documentary, and the right choice turns on who it is for, how far it has to travel, and what you need it to do.
Open the format picker →
The Usual Subjects
Measurement and metrics
- Building performance measures that hold up
- When composite scores help and when they mislead
- Communicating confidence and uncertainty to non-technical stakeholders
- The politics of how organizational metrics get designed
Data visualization
- Choosing chart types that match your argument
- Designing for uncertainty, not false precision
- Interactive vs. static, and when to use each
- Common visualization mistakes in stakeholder reporting