Research with students: what is open, what it takes, and what you get.
Most of my projects are already written and under review, and a student cannot earn a place on a finished paper. What is open is the work that is still unfinished, where your judgment changes the result. Two projects are open this term, both on public data, so you can start the week you write to me.
Staffing Patterns and Why Clinicians Leave
A facility short on nurses fills the gap with agency staff. The cost shows up on one budget line; what it does to continuity, to the people who stayed, and to whether they are still there in six months does not. Federal payroll-based staffing records make daily staffing observable for thousands of facilities. The panel that links staffing patterns at one level to turnover at another does not exist yet. Building it is the project.
What you do
- Assemble the panel from CMS Payroll-Based Journal files, merged across quarters and facility identifiers
- Construct the core measures: agency share, turnover, hours per resident day, staffing volatility
- Descriptives first, then fixed-effects specifications
- Write the codebook that lets a stranger rebuild the dataset
Prerequisites
- One regression course: coefficients, standard errors, interactions
- Stata or R at a working level, or a serious start on the Stata or R guide
- Recommended: the fixed-effects and DiD chapters of the Causal Inference Handbook
- Not required: healthcare knowledge, prior panel-data experience
Turn a Finished Finding Into the Next Teaching Simulator
Every simulator on my teaching page began as a research finding too abstract to survive a lecture. Someone decided which parameters a student controls, which detail to drop, and what has to happen on screen for the insight to land. That translation shows quickly whether you understand a result or only recognize it. The artifact you build is used by the next class, in public, with your name on it.
What you do
- Pick one finding from the published or submitted work and isolate the mechanism a student must feel
- Design the interaction: exposed parameters, the default case, where the surprise happens
- Build it as a self-contained web simulator or notebook in the style of the existing set
- Test it on real students and rebuild the part that failed
Prerequisites
- The ability to explain something clearly; this one is not negotiable
- Intermediate HTML and JavaScript, or Python with a notebook
- Helpful: any tutoring or teaching experience
- Not required: research experience, design training
The chemotherapy-scheduling (deep reinforcement learning) and Ethiopian smallholder projects are now written up, so they are closed for this term. Follow-on work may reopen them; this page will say so when it does.
Three ways in, each scoped to a single term. The hours are a planning guide; credit follows what you contributed, not the level you signed up at.
| Level | Time | What you do | What you leave with |
|---|---|---|---|
| 1 · Explorer | 1–2 hrs/wk · 3–4 weeks | Read one paper and one it argues with; run the matching simulator or open the data; write one page on what you would test next. | Written feedback and an honest answer to whether research suits you. Leaving costs nothing. |
| 2 · Contributor | 2–4 hrs/wk · one term | Own a defined, checkable task (cleaning, a codebook, a constructed measure) with a script that runs top to bottom; biweekly 30-minute check-in. | A specific reference letter, a portfolio artifact, and a named acknowledgment, or co-authorship if your contribution reaches that bar. |
| 3 · Co-author | 4–8 hrs/wk · one term | Contribute to design, data, or analysis and write part of the paper; respond to referees with me. | Co-authorship. At 6–8 hours with a question you brought yourself, shared first authorship with an equal-contribution footnote. |
Four weeks at one to two hours each, using materials already on this site. You can do this before you ever email me, and the week-four output is the strongest application you can send.
Authorship
Co-authorship starts at roughly four hours a week over a term: contribute to design, data, or analysis, write part of the paper, and be able to defend it. Shared first authorship is available at six to eight hours when the question is yours. Author order is discussed before the work starts. Contributions below the bar are acknowledged by name.
Data rules
- Restricted data never leaves approved storage: no personal laptops, no cloud drives
- Human-subjects work requires CITI training before any access
- Public data is the default entry so you can start immediately
- If unsure whether something is permitted, ask first
Reproducibility and AI tools
Every task ends in a script that runs from raw data to output without manual steps. AI tools are allowed on two conditions: you disclose where they were used, and you can defend every line as if you wrote it.
What I commit to
- A reply to any email that names a project
- A scheduled check-in with written feedback on what you submit
- A specific reference letter, or an early, direct conversation if I cannot write a strong one
- Telling you quickly when a direction is not working
Ready to write?
If you have your own question, lead with it. Questions you bring are the route to shared first authorship, and they only need to be close enough to healthcare operations, workforce, or measurement for me to supervise well. Credit options: independent study, honors thesis or capstone, a paid assistantship when funding allows, or portfolio only.
- I have never coded. Is there anything here for me?
- Not this term; both open projects need Stata/R or Python. The four-week preparation above takes you from nothing to a running script with free materials on this site. Do the first three weeks, then write to me.
- Do I need to be a supply chain major?
- No. Students from economics, statistics, public health, computer science, nursing, psychology, and business have all fit. What matters is whether the question holds your attention when the work gets tedious, because it will.
- Will I get paid?
- Sometimes, when grant funding covers it, and I will tell you plainly whether it does when you ask. Most Level 1 and 2 positions are for credit or for the portfolio.
- Not a Rutgers student?
- Both open projects run on public data and work remotely. Projects with restricted or clinical data require affiliation and institutional training.
- Will this help a PhD or medical school application?
- In the ways committees can verify: a letter describing specific work, a writing sample, a reproducible analysis, and from Level 3 a manuscript. It does not substitute for grades or scores.