To illustrate just how hard: much of my work in academia was assisting in designing experiments or analyzing data for other people working on PhDs in "how to design decent tests" (i.e., Instructional Psychology). There was an entire department of the university dedicated to studying just that problem.