Almost every business question is answered using a sample a subset of customers, employees, transactions, or events that stands in for a larger population we cannot fully observe. Whether the sample reasonably represents the population determines whether the answer is informative or actively misleading. When the sample is biased, no amount of statistical
sophistication applied afterward can fully recover the truth. Garbage in, garbage out, as the saying goes, but the garbage is often invisible because the sampling step happens before the data even arrives.
