The most important skill for an empirical researcher is not math, but judgment.