Statistical sampling is the art of selecting a sample that embodies the essence of the population you wish to study. A good sample, therefore, is a miniature version of the population.
Serving this noble goal is an extravagant assortment of sampling methods.
Some, like probability proportional to size sampling, and cluster sampling, and to some extent quota sampling, are designed to come to your rescue when the disquieting realization dawns that your funder’s desire to obtain a magnificently representative sample is somehow completely detached from their ability to pay for it.
Then there are others— convenience sampling and haphazard sampling to name two— that are named in that special way that wraps unsettlingly unscientific methods in the euphemistic cloak of false rigor and respectability.
Over the years, people and businesses have also crafted personalized notions of what “sampling” means, leading to creative interpretations.
Consider, for instance, the defiant nonchalance of Frank Barone from the TV sitcom Everybody Loves Raymond (S07E12, “Grandpa Steals”) as he helps himself to fistfuls of trail mix at the supermarket with the firm belief that he is only “sampling”.