> They just run regression on it and anything with p<00.5 that seemed fun was published. How do you think science operates in actual practice? OkCupid at least

> They just run regression on it and anything with p<00.5 that seemed fun was published. How do you think science operates in actual practice? OkCupid at least has a sample large enough that you can commit all sorts of statistical sins and still produce halfway decent conclusions, unlike your run-off-the-mill n=10 study.