Measurements are also only as good as your experimental design, your data collection tools, your methodology, and your statistical analysis.

Which is my way of saying, "Doing good science is actually pretty hard."