You do not need a grant, an ethics committee or a hospital ward to publish a real epidemiological study. You need a question, a laptop and about forty hours with a database the CDC has been quietly maintaining since 1999.

Ninety per cent of a clean analysis is clean data entry. Here is how to build a dataset in SPSS that will still make sense to you six months later, and how to run the four tests a cross-sectional paper actually needs.

A question you can answer beats a question you admire. Most student projects fail at the question, not at the analysis — and the failure is visible on day one if you know what to look for.