I'm a quantitative methodologist at the University of Nebraska–Lincoln. I help people work out whether their data measures what they think it does, and when the same problem keeps coming up, I build an R or Shiny tool for it.
Most of it starts the same way: a dataset somebody needs to trust and can't yet.
- recode-studio: clean messy string variables without writing R by hand. Exports the recode rules, so next quarter's data gets the same treatment as last quarter's.
- table-explorer: point it at files or a database and get auto-detected primary and foreign keys, an interactive ERD, and a report you can hand to someone. R/Shiny, with a Python/Streamlit port.
- lspc-weather: skydiving weather for my drop zone. Every flag cites the published USPA, FAA, or club rule behind it, and where nothing published sets a threshold, it raises no flag at all.
- mplusplus: open Mplus
.dgmdiagrams and.outmodel output in the browser and turn them into SVG path diagrams you can rearrange and restyle, with significance marked from the output. Runs from a local file, no server. - pleasantry: a toy project started in a workshop at posit::conf(2026). Randomized compliments and farewells, from sincere through backhanded to pseudo-Shakespearean.