Mistake Master
A Sampling Method Is the Set of Samples It Can Produce
You'll learnwhat separates the named sampling designs, by running each one forty times on the same school and measuring how far its estimates scatter.
Twelve homerooms of twenty students, three per grade, and 42.5% of the school drives in. Take 60 students four different ways. A simple random sample scatters its estimates over a range of 21.7 points across forty draws. Stratifying — an SRS inside every grade — narrows that to 16.7, and every single stratified sample has exactly 15 students from each grade, because that was decided by the design rather than by chance. Take three whole homerooms instead and the range explodes to 60 points, because a homeroom is one grade and a draw can miss two grades completely. Take every fourth name on the roster and there are only four samples in existence. The methods differ in which samples they can produce at all, and that is the whole of what separates them.