Claude and I

All analysis conducted in this project was done with assistance from Claude Code.

Claude was most useful in three areas. Web scraping: knowing the principles is one thing; implementing the routines reliably is another, and Claude handled the bulk of that implementation — although we are still figuring out how to get past the tests that ask us to prove we are human, which feels increasingly ironic. Data visualisation: once given the right instructions, Claude delivered plot functions quickly and was precise in axis labelling, formatting, and all the detail work that a perfectionist can spend hours on. I meant a perfectionist. Coding output: Claude formatted statistical output, refactored exploratory cell code into reusable functions in the project’s .py libraries, and kept the codebase cleaner than it would otherwise have been.

The statistical plan, the choice of methods, the evaluation of results, and the narrative are mine. If something is incorrect, that is on me. Claude was also used to double-check scientific accuracy and to keep the writing from going off track.

As for the editorial relationship: Claude was asked to review every joke I tried to land. We disagreed regularly. We are now seeking couple therapy.