This only works if you're measuring something that has a symmetric, center-weighted distribution (e.g. a bell curve). If you're measuring something flat, spiky or asymmetric, saturation is not the right strategy. Don't fall into this research trap.