Keeping a research log

Of all the small habits I picked up in the PhD, the one I would recommend first is a research log: a plain, dated record of what I did and why.

It is not a diary and not a polished notebook. Each entry is short, usually a few lines:

• Date.
• What I did. Ran a model, cleaned a dataset, read three papers, rewrote a section.
• What happened. Results, errors, surprises.
• What I decided, and the reason.
• Next step.

Why it helps:

Decisions are easy to forget. Months later, a reviewer or supervisor asks why a variable was dropped or why a sample starts in a certain year. Without a log, the answer is a guess. With one, it is a search.

Analysis becomes reproducible. In DEA and similar methods, small choices change results: which inputs and outputs, which model, how outliers were handled. Writing down each version, with file names and settings, makes it possible to rebuild a result exactly.

It shows progress on slow days. Research often feels like nothing is moving. Looking back over a month of entries shows that it was.

It makes writing easier. Much of a methods section is already in the log, in rough form.

A few practical choices:

• Keep it in one place. A single text file, a notes app or a paper notebook all work. Consistency matters more than the tool.
• Write it at the end of each working session, while details are fresh.
• Link to files and code versions instead of describing them from memory.
• Record failures too. Approaches that did not work are some of the most useful entries later.

It takes five minutes a day, and it has saved me hours more than once.

---

Photos: promofocus on Pixabay, used under the Pixabay Content License. These are not my photographs.